US20230237102A1

US20230237102A1 - Transparent referrals for distributed file servers

Info

Publication number: US20230237102A1
Application number: US18/069,920
Authority: US
Inventors: Durga Mahesh Arikatla; Hemanth Kumar Thummala; Manish Gupta; Manoj Naik; Saji Kumar Vijaya Kumari Rajendran Nair; Shyamsunder Prayagchand Rathi; Sukruth Sridharan
Original assignee: Nutanix Inc
Current assignee: Nutanix Inc
Priority date: 2016-12-02
Filing date: 2022-12-21
Publication date: 2023-07-27
Also published as: US20180157752A1; US11562034B2

Abstract

Examples of systems described herein include a virtualized file server including a first file server virtual machine and a second file server virtual machine configured to manage a distributed file share of storage items. The second file server virtual machine is configured to manage a particular storage item of the distributed file share of storage items. The first file server virtual machine is configured to, in response to receipt of a referral request for a file share path for the particular storage item from a client, look up a file share path for the particular storage item in a map of at least a portion of the distributed file share of storage items, and provide a referral with the file share path that identifies the second file server virtual machine.

Description

CROSS REFERENCE TO RELATED APPLICATION(S)

This application is a continuation of U.S. application Ser. No. 15/829,731 filed Dec. 1, 2017, which claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/429,638 entitled “TRANSPARENT REFERRALS FOR A DISTRIBUTED FILE SERVER”, filed Dec. 2, 2016, and also claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/429,656 entitled “RECOVERY OPTIMIZATION”, filed Dec. 2, 2016, and also claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/429,992 entitled “HOME DIRECTORY MOBILITY FOR DISTRIBUTED FILE SERVERS”, filed Dec. 5, 2016, and also claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/430,865 entitled “EFFICIENT ENUMERATION FOR VIRTUALIZED FILE SERVERS”, filed Dec. 6, 2016, and also claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/430,550 entitled “LOCAL DOMAIN CONTROLLER DISCOVERY”, filed Dec. 6, 2016, and also claims the benefit under 35 U.S.C. 119 of the earlier filing date of U.S. Provisional Application No. 62/431,798 entitled “REFERRAL PATHS FOR VIRTUALIZED FILE SYSTEMS”, filed Dec. 8, 2016. These aforementioned applications are hereby incorporated by reference in their entirety, for any purpose.

TECHNICAL FIELD

Examples described herein pertain to distributed and cloud computing systems. Examples of distributed virtual file server systems are described.

BACKGROUND

A “virtual machine” or a “VM” refers to a specific software-based implementation of a machine in a virtualization environment, in which the hardware resources of a real computer (e.g., CPU, memory, etc.) are virtualized or transformed into the underlying support for the fully functional virtual machine that can run its own operating system and applications on the underlying physical resources just like a real computer.
Virtualization generally works by inserting a thin layer of software directly on the computer hardware or on a host operating system. This layer of software contains a virtual machine monitor or “hypervisor” that allocates hardware resources dynamically and transparently. Multiple operating systems run concurrently on a single physical computer and share hardware resources with each other. By encapsulating an entire machine, including CPU, memory, operating system, and network devices, a virtual machine is completely compatible with most standard operating systems, applications, and device drivers. Most modern implementations allow several operating systems and applications to safely run at the same time on a single computer, with each having access to the resources it needs when it needs them.
Virtualization allows one to run multiple virtual machines on a single physical machine, with each virtual machine sharing the resources of that one physical computer across multiple environments. Different virtual machines can run different operating systems and multiple applications on the same physical computer.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A illustrates a clustered virtualization environment according to some particular embodiments.

FIG. 1B illustrates data flow within an example clustered virtualization environment according to particular embodiments.

FIG. 2A illustrates a clustered virtualization environment implementing a virtualized file server (VFS) 202 according to particular embodiments.

FIG. 2B illustrates data flow within a clustered virtualization environment implementing a VFS instance 202 in which stored items such as files and folders used by user VMs are stored locally on the same host machines as the user VMs according to particular embodiments.

FIG. 3A illustrates an example hierarchical structure of a VFS instance in a cluster according to particular embodiments.

FIG. 3B illustrates two example host machines, each providing file storage services for portions of two VFS instances FS1 and FS2 according to particular embodiments.

FIG. 3C illustrates example interactions between a client and host machines on which different portions of a VFS instance are stored according to particular embodiments.

FIG. 3D illustrates an example virtualized file server having a failover capability according to particular embodiments.

FIG. 3E illustrates an example virtualized file server that has recovered from a failure of Controller/Service VM CVM by switching to an alternate Controller/Service VM CVM according to particular embodiments.

FIG. 3F illustrates an example virtualized file server that has recovered from failure of a FSVM by electing a new leader FSVM according to particular embodiments.

FIGS. 3G and 3H illustrate example virtualized file servers that have recovered from failure of a host machine by switching to another Controller/Service VM and another FSVM according to particular embodiments.

FIGS. 4A and 4B illustrate an example hierarchical namespace of a file server according to particular embodiments.

FIG. 4C illustrates distribution of stored data amongst host machines in a virtualized file server according to particular embodiments.

FIG. 5 illustrates an example method for accessing data in a virtualized file server according to particular embodiments.

FIG. 6 illustrates an example of how a file server ‘FS1’ may be deployed across multiple clusters according to particular embodiments.

FIG. 7 is metadata database management for a distributed file system according to particular embodiments.

FIG. 8 illustrates example interactions between a client and host machines on which different portions of a VFS instance are stored according to particular embodiments

DETAILED DESCRIPTION

One reason for the broad adoption of virtualization in modern business and computing environments is because of the resource utilization advantages provided by virtual machines. Without virtualization, if a physical machine is limited to a single dedicated operating system, then during periods of inactivity by the dedicated operating system the physical machine is not utilized to perform useful work. This is wasteful and inefficient if there are users on other physical machines which are currently waiting for computing resources. To address this problem, virtualization allows multiple VMs to share the underlying physical resources so that during periods of inactivity by one VM, other VMs can take advantage of the resource availability to process workloads. This can produce great efficiencies for the utilization of physical devices, and can result in reduced redundancies and better resource cost management.
Furthermore, there are now products that can aggregate multiple physical machines, running virtualization environments to not only utilize the processing power of the physical devices to aggregate the storage of the individual physical devices to create a logical storage pool wherein the data may be distributed across the physical devices but appears to the virtual machines to be part of the system that the virtual machine is hosted on. Such systems operate under the covers by using metadata, which may be distributed and replicated any number of times across the system, to locate the indicated data. These systems are commonly referred to as clustered systems, wherein the resources of the group are pooled to provide logically combined, but physically separate systems.
Particular embodiments provide an architecture for implementing virtualized file servers in a virtualization environment. In particular embodiments, a virtualized file server may include a set of File Server Virtual Machines (VMs) that execute on host machines and process storage access operations requested by user VMs executing on the host machines. The file server VMs may communicate with storage controllers provided by Controller/Service VMs executing on the host machines to store and retrieve storage items, such as files and folders, on storage devices associated with the host machines. The storage items may be distributed amongst multiple host machines. The file server VMs may maintain a storage map, such as a sharding map, that maps names or identifiers of storage items, such as folders, files, or portions thereof, to their locations. When a user application executing in a user VM on one of the host machines initiates a storage access operation, such as reading or writing data from or to a storage item or modifying metadata associated with the storage item, the user VM may send the storage access operation in a request to one of the file server VMs on one of the host machines. In particular embodiments, a file server VM executing on a host machine that receives a storage access request may use the storage map to determine whether the requested storage item is located on the host machine (or otherwise associated with the file server VM or Controller/Service VM on the host machine). If so, the file server VM executes the requested operation. Otherwise, the file server VM responds to the request with an indication that the requested storage item is not on the host machine, and may redirect the requesting user VM to the host machine on which the storage map indicates the storage item is located. The client may cache the address of the host machine on which the storage item is located, so that the client may send subsequent requests for the storage item directly to that host machine.
In particular embodiments, the virtualized file server determines the location, e.g., host machine, at which to store a storage item such as a file or folder when the storage item is created. A file server VM may attempt to create a file or folder using a Controller/Service VM on the same host machine as the user VM that requested creation of the file, so that the Controller/Service VM that controls access operations to the storage item is co-located with the requesting user VM. In this way, file access operations between the user VM that is known to be associated with the storage item and is thus likely to access the storage item again (e.g., in the near future and/or on behalf of the same user) may use local communication or short-distance communication to improve performance, e.g., by reducing access times or increasing access throughput. Further, the virtualized file server may also attempt to store the storage item on a storage device that is local to the Controller/Service VM being used to create the storage item, so that storage access operations between the Controller/Service VM and the storage device may use local or short-distance communication.
Further details of aspects, objects, and advantages of the invention are described below in the detailed description, drawings, and claims. Both the foregoing general description and the following detailed description are exemplary and explanatory, and are not intended to be limiting as to the scope of the invention. Particular embodiments may include all, some, or none of the components, elements, features, functions, operations, or steps of the embodiments disclosed above. The subject matter which can be claimed comprises not only the combinations of features as set out in the attached claims but also any other combination of features in the claims, wherein each feature mentioned in the claims can be combined with any other feature or combination of other features in the claims. Furthermore, any of the embodiments and features described or depicted herein can be claimed in a separate claim and/or in any combination with any embodiment or feature described or depicted herein or with any of the features of the attached claims.
FIG. 1A illustrates a clustered virtualization environment according to some particular embodiments. The architecture of FIG. 1A can be implemented for a distributed platform that contains multiple host machines 100 a-c that manage multiple tiers of storage. The multiple tiers of storage may include network-attached storage (NAS) that is accessible through network 140, such as, by way of example and not limitation, cloud storage 126, which may be accessible through the Internet, or local network-accessible storage 128 (e.g., a storage area network (SAN)). Unlike the prior art, the present embodiment also permits local storage 122 that is within or directly attached to the server and/or appliance to be managed as part of storage pool 160. Examples of such storage include Solid State Drives 125 (henceforth “SSDs”), Hard Disk Drives 127 (henceforth “HDDs” or “spindle drives”), optical disk drives, external drives (e.g., a storage device connected to a host machine via a native drive interface or a direct attach serial interface), or any other directly attached storage. These collected storage devices, both local and networked, form storage pool 160. Virtual disks (or “vDisks”) can be structured from the storage devices in storage pool 160, as described in more detail below. As used herein, the term vDisk refers to the storage abstraction that is exposed by a Controller/Service VM (CVM) to be used by a user VM. In some embodiments, the vDisk is exposed via iSCSI (“internet small computer system interface”) or NFS (“network file system”) and is mounted as a virtual disk on the user VM.
Each host machine 100 a-c runs virtualization software, such as VMWARE ESX(I), MICROSOFT HYPER-V, or REDHAT KVM. The virtualization software includes hypervisor 130 a-c to manage the interactions between the underlying hardware and the one or more user VMs 101 a, 102 a, 101 b, 102 b, 101 c, and 102 c that run client software. Though not depicted in FIG. 1A, a hypervisor may connect to network 140. In particular embodiments, a host machine 100 may be a physical hardware computing device; in particular embodiments, a host machine 100 may be a virtual machine.
CVMs 110 a-c are used to manage storage and input/output (“I/O”) activities according to particular embodiments. These special VMs act as the storage controller in the currently described architecture. Multiple such storage controllers may coordinate within a cluster to form a unified storage controller system. CVMs 110 may run as virtual machines on the various host machines 100, and work together to form a distributed system 110 that manages all the storage resources, including local storage 122, networked storage 128, and cloud storage 126. The CVMs may connect to network 140 directly, or via a hypervisor. Since the CVMs run independent of hypervisors 130 a-c, this means that the current approach can be used and implemented within any virtual machine architecture, since the CVMs can be used in conjunction with any hypervisor from any virtualization vendor.
A host machine may be designated as a leader node within a cluster of host machines. For example, host machine 100 b, as indicated by the asterisks, may be a leader node. A leader node may have a software component designated to perform operations of the leader. For example, CVM 110 b on host machine 100 b may be designated to perform such operations. A leader may be responsible for monitoring or handling requests from other host machines or software components on other host machines throughout the virtualized environment. If a leader fails, a new leader may be designated. In particular embodiments, a management module (e.g., in the form of an agent) may be running on the leader node.
Each CVM 110 a-c exports one or more block devices or NFS server targets that appear as disks to user VMs 101 and 102. These disks are virtual, since they are implemented by the software running inside CVMs 110 a-c. Thus, to user VMs 101 and 102, CVMs 110 a-c appear to be exporting a clustered storage appliance that contains some disks. All user data (including the operating system) in the user VMs 101 and 102 reside on these virtual disks.
Significant performance advantages can be gained by allowing the virtualization system to access and utilize local storage 122 as disclosed herein. This is because I/O performance is typically much faster when performing access to local storage 122 as compared to performing access to networked storage 128 across a network 140. This faster performance for locally attached storage 122 can be increased even further by using certain types of optimized local storage devices, such as SSDs. Further details regarding methods and mechanisms for implementing the virtualization environment illustrated in FIG. 1A are described in U.S. Pat. No. 8,601,473, which is hereby incorporated by reference in its entirety.
FIG. 1B illustrates data flow within an example clustered virtualization environment according to particular embodiments. As described above, one or more user VMs and a CVM may run on each host machine 100 along with a hypervisor. As a user VM performs I/O operations (e.g., a read operation or a write operation), the I/O commands of the user VM may be sent to the hypervisor that shares the same server as the user VM. For example, the hypervisor may present to the virtual machines an emulated storage controller, receive an I/O command and facilitate the performance of the I/O command (e.g., via interfacing with storage that is the object of the command, or passing the command to a service that will perform the I/O command). An emulated storage controller may facilitate I/O operations between a user VM and a vDisk. A vDisk may present to a user VM as one or more discrete storage drives, but each vDisk may correspond to any part of one or more drives within storage pool 160. Additionally or alternatively, Controller/Service VM 110 a-c may present an emulated storage controller either to the hypervisor or to user VMs to facilitate I/O operations. CVMs 110 a-c may be connected to storage within storage pool 160. CVM 110 a may have the ability to perform I/O operations using local storage 122 a within the same host machine 100 a, by connecting via network 140 to cloud storage 126 or networked storage 128, or by connecting via network 140 to local storage 122 b-c within another host machine 100 b-c (e.g., via connecting to another CVM 110 b or 110 c). In particular embodiments, any suitable computing system 700 may be used to implement a host machine 100.
File System Architecture
FIG. 2A illustrates a clustered virtualization environment implementing a virtualized file server (VFS) 202 according to particular embodiments. In particular embodiments, the VFS 202 provides file services to user virtual machines (user VMs) 101 and 102. The file services may include storing and retrieving data persistently, reliably, and efficiently. The user virtual machines 101 and 102 may execute user processes, such as office applications or the like, on host machines 200 a-c. The stored data may be represented as a set of storage items, such as files organized in a hierarchical structure of folders (also known as directories), which can contain files and other folders.
In particular embodiments, the VFS 202 may include a set of File Server Virtual Machines (FSVMs) 170 a-c that execute on host machines 200 a-c and process storage item access operations requested by user VMs 200 a-c executing on the host machines 200 a-c. The FSVMs 170 a-c may communicate with storage controllers provided by CVMs 110 a-c executing on the host machines 200 a-c to store and retrieve files, folders, or other storage items on local storage 122 a-c associated with, e.g., local to, the host machines 200 a-c. The network protocol used for communication between user VMs 101 and 102, FSVMs 170 a-c, and CVMs 110 a-c via the network 140 may be Internet Small Computer Systems Interface (iSCSI), Server Message Block (SMB), Network File System (NFS), pNFS (Parallel NFS), or another appropriate protocol.
For the purposes of VFS 202, host machine 200 c may be designated as a leader node within a cluster of host machines. In this case, FSVM 170 c on host machine 100 c may be designated to perform such operations. A leader may be responsible for monitoring or handling requests from FSVMs on other host machines throughout the virtualized environment. If FSVM 170 c fails, a new leader may be designated for VFS 202.
In particular embodiments, the user VMs 101 and 102 may send data to the VFS 202 using write requests, and may receive data from it using read requests. The read and write requests, and their associated parameters, data, and results, may be sent between a user VM 101 a and one or more file server VMs (FSVMs) 170 a-c located on the same host machine as the user VM 101 a or on different host machines from the user VM 101 a. The read and write requests may be sent between host machines 200 a-c via network 140, e.g., using a network communication protocol such as iSCSI, CIFS, SMB, TCP, IP, or the like. When a read or write request is sent between two VMs located on the same one of the host machines 200 a-c (e.g., between the user VM 101 a and the FSVM 170 a located on the host machine 200 a), the request may be sent using local communication within the host machine 200 a instead of via the network 140. As described above, such local communication may be substantially faster than communication via the network 140. The local communication may be performed by, e.g., writing to and reading from shared memory accessible by the user VM 101 a and the FSVM 170 a, sending and receiving data via a local “loopback” network interface, local stream communication, or the like.
In particular embodiments, the storage items stored by the VFS 202, such as files and folders, may be distributed amongst multiple host machines 200 a-c. In particular embodiments, when storage access requests are received from the user VMs 101 and 102, the VFS 202 identifies host machines 200 a-c at which requested storage items, e.g., folders, files, or portions thereof, are stored, and directs the user VMs 101 and 102 to the locations of the storage items. The FSVMs 170 may maintain a storage map, such as a sharding map 360 (shown in FIG. 3C), that maps names or identifiers of storage items to their corresponding locations. The storage map may be a distributed data structure of which copies are maintained at each FSVM 170 a-c and accessed using distributed locks or other storage item access operations. Alternatively, the storage map may be maintained by a leader node such as the host machine 200 c, and the other host machines 200 a and 200 b may send requests to query and update the storage map to the leader host machine 200 c. Other implementations of the storage map are possible using appropriate techniques to provide asynchronous data access to a shared resource by multiple readers and writers. The storage map may map names or identifiers of storage items in the form of text strings or numeric identifiers, such as folder names, files names, and/or identifiers of portions of folders or files (e.g., numeric start offset positions and counts in bytes or other units) to locations of the files, folders, or portions thereof. Locations may be represented as names of FSVMs 170 a-c, e.g., “FSVM-1”, as network addresses of host machines 200 a-c on which FSVMs 170 a-c are located (e.g., “ip-addr1” or 128.1.1.10), or as other types of location identifiers.
When a user application executing in a user VM 101 a on one of the host machines 200 a initiates a storage access operation, such as reading or writing data, the user VM 101 a may send the storage access operation in a request to one of the FSVMs 170 a-c on one of the host machines 200 a-c. A FSVM executing on a host machine 200 b that receives a storage access request may use the storage map to determine whether the requested file or folder is located on the host machine 200 b (or otherwise associated with the FSVM 170 b or Controller/Service VM 110 b on the host machine 200 b). If the requested file or folder is located on the host machine 200 b (or otherwise associated with a VM on it), the FSVM 170 b executes the requested storage access operation. Otherwise, the FSVM 170 b responds to the request with an indication that the data is not on the host machine 200 b, and may redirect the requesting user VM 101 a to the host machine 200 c on which the storage map indicates the file or folder is located. The client may cache the address of the host machine 200 c on which the file or folder is located, so that it may send subsequent requests for the file or folder directly to the host machine 200 c.
As an example and not by way of limitation, the location of a file or a folder may be pinned to a particular host machine 200 a by sending a file service operation that creates the file or folder to a CVM 110 a located on the particular host machine 200 a. The CVM 110 a subsequently processes file service commands for that file and sends corresponding storage access operations to storage devices associated with the file. The CVM 110 a may associate local storage 122 a with the file if there is sufficient free space on local storage 122 a. Alternatively, the CVM 110 a may associate a storage device located on another host machine 200 b, e.g., in local storage 122 b, with the file under certain conditions, e.g., if there is insufficient free space on the local storage 122 a, or if storage access operations between the CVM 110 a and the file are expected to be infrequent. Files and folders, or portions thereof, may also be stored on other storage devices, such as the network-attached storage (NAS) 128 or the cloud storage 126 of the storage pool 160.
In particular embodiments, a name service 220, such as that specified by the Domain Name System (DNS) Internet protocol, may communicate with the host machines 200 a-c via the network 140 and may store a database of domain name (e.g., host name) to IP address mappings. The name service 220 may be queried by the User VMs 101 to determine the IP address of a particular host machine 200 a-c given a name of the host machine, e.g., to determine the IP address of the host name ip-addr1 for the host machine 200 a. The name service 220 may be located on a separate server computer system or on one or more of the host machines 200. The names and IP addresses of the host machines of the VFS instance 202, e.g., the host machines 200, may be stored in the name service 220 so that the user VMs 101 may determine the IP address of each of the host machines 200. The name of each VFS instance 202, e.g., FS1, FS2, or the like, may be stored in the name service 220 in association with a set of one or more names that contains the name(s) of the host machines 200 of the VFS instance 202. For example, the file server instance name FS1.domain.com may be associated with the host names ip-addr1, ip-addr2, and ip-addr3 in the name service 220, so that a query of the name service 220 for the server instance name “FS1” or “FS1.domain.com” returns the names ip-addr1, ip-addr2, and ip-addr3. Further, the name service 220 may return the names in a different order for each name lookup request, e.g., using round-robin ordering, so that the sequence of names (or addresses) returned by the name service for a file server instance name is a different permutation for each query until all the permutations have been returned in response to requests, at which point the permutation cycle starts again, e.g., with the first permutation. In this way, storage access requests from user VMs 101 may be balanced across the host machines 200, since the user VMs 101 submit requests to the name service 220 for the address of the VFS instance 202 for storage items for which the user VMs 101 do not have a record or cache entry, as described below.
In particular embodiments, each FSVM 170 may have two IP addresses: an external IP address and an internal IP address. The external IP addresses may be used by SMB/CIFS clients, such as user VMs 101, to connect to the FSVMs 170. The external IP addresses may be stored in the name service 220. The IP addresses ip-addr1, ip-addr2, and ip-addr3 described above are examples of external IP addresses. The internal IP addresses may be used for iSCSI communication to CVMs 110, e.g., between the FSVMs 170 and the CVMs 110, and for communication between the CVMs 110 and storage devices in the storage pool 160. Other internal communications may be sent via the internal IP addresses as well, e.g., file server configuration information may be sent from the CVMs 110 to the FSVMs 170 using the internal IP addresses, and the CVMs 110 may get file server statistics from the FSVMs 170 via internal communication as needed.
Since the VFS 202 is provided by a distributed set of FSVMs 170 a-c, the user VMs 101 and 102 that access particular requested storage items, such as files or folders, do not necessarily know the locations of the requested storage items when the request is received. A distributed file system protocol, e.g., MICROSOFT DFS or the like, is therefore used, in which a user VM 101 a may request the addresses of FSVMs 170 a-c from a name service 220 (e.g., DNS). The name service may send one or more network addresses of FSVMs 170 a-c to the user VM 101 a, in an order that changes for each subsequent request. These network addresses are not necessarily the addresses of the FSVM 170 b on which the storage item requested by the user VM 101 a is located, since the name service 220 does not necessarily have information about the mapping between storage items and FSVMs 170 a-c. Next, the user VM 170 a may send an access request to one of the network addresses provided by the name service, e.g., the address of FSVM 170 b. The FSVM 170 b may receive the access request and determine whether the storage item identified by the request is located on the FSVM 170 b. If so, the FSVM 170 b may process the request and send the results to the requesting user VM 101 a. However, if the identified storage item is located on a different FSVM 170 c, then the FSVM 170 b may redirect the user VM 101 a to the FSVM 170 c on which the requested storage item is located by sending a “redirect” response referencing FSVM 170 c to the user VM 101 a. The user VM 101 a may then send the access request to FSVM 170 c, which may perform the requested operation for the identified storage item.
A particular VFS 202, including the items it stores, e.g., files and folders, may be referred to herein as a VFS “instance” 202 and may have an associated name, e.g., FS1, as described above. Although a VFS instance 202 may have multiple FSVMs distributed across different host machines 200, with different files being stored on different host machines 200, the VFS instance 202 may present a single name space to its clients such as the user VMs 101. The single name space may include, for example, a set of named “shares” and each share may have an associated folder hierarchy in which files are stored. Storage items such as files and folders may have associated names and metadata such as permissions, access control information, size quota limits, file types, files sizes, and so on. As another example, the name space may be a single folder hierarchy, e.g., a single root directory that contains files and other folders. User VMs 101 may access the data stored on a distributed VFS instance 202 via storage access operations, such as operations to list folders and files in a specified folder, create a new file or folder, open an existing file for reading or writing, and read data from or write data to a file, as well as storage item manipulation operations to rename, delete, copy, or get details, such as metadata, of files or folders. Note that folders may also be referred to herein as “directories.”
In particular embodiments, storage items such as files and folders in a file server namespace may be accessed by clients such as user VMs 101 by name, e.g., “\Folder-1\File-1” and “\Folder-2 \File-2” for two different files named File-1 and File-2 in the folders Folder-1 and Folder-2, respectively (where Folder-1 and Folder-2 are sub-folders of the root folder). Names that identify files in the namespace using folder names and file names may be referred to as “path names.” Client systems may access the storage items stored on the VFS instance 202 by specifying the file names or path names, e.g., the path name “\Folder-1 \File-1”, in storage access operations. If the storage items are stored on a share (e.g., a shared drive), then the share name may be used to access the storage items, e.g., via the path name “\\Share-1\Folder-1 \File-1” to access File-1 in folder Folder-1 on a share named Share-1.
In particular embodiments, although the VFS instance 202 may store different folders, files, or portions thereof at different locations, e.g., on different host machines 200, the use of different host machines or other elements of storage pool 160 to store the folders and files may be hidden from the accessing clients. The share name is not necessarily a name of a location such as a host machine 200. For example, the name Share-1 does not identify a particular host machine 200 on which storage items of the share are located. The share Share-1 may have portions of storage items stored on three host machines 200 a-c, but a user may simply access Share-1, e.g., by mapping Share-1 to a client computer, to gain access to the storage items on Share-1 as if they were located on the client computer. Names of storage items, such as file names and folder names, are similarly location-independent. Thus, although storage items, such as files and their containing folders and shares, may be stored at different locations, such as different host machines 200 a-c, the files may be accessed in a location-transparent manner by clients (such as the user VMs 101 and 102). Thus, users at client systems need not specify or know the locations of each storage item being accessed. The VFS 202 may automatically map the file names, folder names, or full path names to the locations at which the storage items are stored. As an example and not by way of limitation, a storage item's physical location may be specified by the name or address of the host machine 200 a-c on which the storage item is located, the name, address, or identity of the FSVM 170 a-c that provides access to the storage item on the host machine 200 a-c on which the storage item is located, the particular device (e.g., SSD or HDD) of the local storage 122 a (or other type of storage in storage pool 160) on which the storage item is located, and the address on the device, e.g., disk block numbers. A storage item such as a file may be divided into multiple parts that may be located on different host machines 200 a-c, in which case access requests for a particular portion of the file may be automatically mapped to the location of the portion of the file based on the portion of the file being accessed (e.g., the offset from the beginning of the file and the number of bytes being accessed).
In particular embodiments, VFS 202 determines the location, e.g., particular host machine 200 a-c, at which to store a storage item when the storage item is created. For example, a FSVM 170 a may attempt to create a file or folder using a Controller/Service VM 110 a on the same host machine 200 a as the user VM 101 a that requested creation of the file, so that the Controller/Service VM 110 a that controls access operations to the file folder is co-located with the user VM 101 a. In this way, since the user VM 101 a is known to be associated with the file or folder and is thus likely to access the file again, e.g., in the near future or on behalf of the same user, access operations may use local communication or short-distance communication to improve performance, e.g., by reducing access times or increasing access throughput. If there is a local CVM 110 a on the same host machine as the FSVM 170 a, the FSVM 170 a may identify it and use it by default. If there is no local CVM 110 a on the same host machine as the FSVM 170 a, a delay may be incurred for communication between the FSVM 170 a and a CVM 110 b on a different host machine 200 b. Further, the VFS 202 may also attempt to store the file on a storage device that is local to the CVM 110 a being used to create the file, such as local storage 122 a, so that storage access operations between the CVM 110 a and local storage 122 a may use local or short-distance communication.
In particular embodiments, if a CVM 110 a is unable to store the storage item in local storage 122 a, e.g., because local storage 122 a does not have sufficient available free space, then the file may be stored in local storage 122 b of a different host machine 200 b. In this case, the stored file is not physically local to the host machine 200 a, but storage access operations for the file are performed by the locally-associated CVM 110 a and FSVM 170 a, and the CVM 110 a may communicate with local storage 122 b on the remote host machine 200 b using a network file sharing protocol, e.g., iSCSI, SAMBA or the like.
In particular embodiments, if a virtual machine, such as a user VM 101 a, CVM 110 a, or FSVM 170 a, moves from a host machine 200 a to a destination host machine 200 b, e.g., because of resource availability changes, and data items such as files or folders associated with the VM are not locally accessible on the destination host machine 200 b, then data migration may be performed for the data items associated with the moved VM to migrate them to the new host machine 200 b, so that they are local to the moved VM on the new host machine 200 b. FSVMs 170 may detect removal and addition of CVMs 110 (as may occur, for example, when a CVM 110 fails or is shut down) via the iSCSI protocol or other technique, such as heartbeat messages. As another example, a FSVM 170 may determine that a particular file's location is to be changed, e.g., because a disk on which the file is stored is becoming full, because changing the file's location is likely to reduce network communication delays and therefore improve performance, or for other reasons. Upon determining that a file is to be moved, VFS 202 may change the location of the file by, for example, copying the file from its existing location(s), such as local storage 122 a of a host machine 200 a, to its new location(s), such as local storage 122 b of host machine 200 b (and to or from other host machines, such as local storage 122 c of host machine 200 c if appropriate), and deleting the file from its existing location(s). Write operations on the file may be blocked or queued while the file is being copied, so that the copy is consistent. The VFS 202 may also redirect storage access requests for the file from an FSVM 170 a at the file's existing location to a FSVM 170 b at the file's new location.
In particular embodiments, VFS 202 includes at least three File Server Virtual Machines (FSVMs) 170 a-c located on three respective host machines 200 a-c. To provide high-availability, there may be a maximum of one FSVM 170 a for a particular VFS instance 202 per host machine 200 in a cluster. If two FSVMs 170 are detected on a single host machine 200, then one of the FSVMs 170 may be moved to another host machine automatically, or the user (e.g., system administrator) may be notified to move the FSVM 170 to another host machine. The user may move a FSVM 170 to another host machine using an administrative interface that provides commands for starting, stopping, and moving FSVMs 170 between host machines 200.
In particular embodiments, two FSVMs 170 of different VFS instances 202 may reside on the same host machine 200 a. If the host machine 200 a fails, the FSVMs 170 on the host machine 200 a become unavailable, at least until the host machine 200 a recovers. Thus, if there is at most one FSVM 170 for each VFS instance 202 on each host machine 200 a, then at most one of the FSVMs 170 may be lost per VFS 202 per failed host machine 200. As an example, if more than one FSVM 170 for a particular VFS instance 202 were to reside on a host machine 200 a, and the VFS instance 202 includes three host machines 200 a-c and three FSVMs 170, then loss of one host machine would result in loss of two-thirds of the FSVMs 170 for the VFS instance 202, which would be more disruptive and more difficult to recover from than loss of one-third of the FSVMs 170 for the VFS instance 202.
In particular embodiments, users, such as system administrators or other users of the user VMs 101, 102, may expand the cluster of FSVMs 170 by adding additional FSVMs 170. Each FSVM 170 a may be associated with at least one network address, such as an IP (Internet Protocol) address of the host machine 200 a on which the FSVM 170 a resides. There may be multiple clusters, and all FSVMs of a particular VFS instance are ordinarily in the same cluster. The VFS instance 202 may be a member of a MICROSOFT ACTIVE DIRECTORY domain, which may provide authentication and other services such as name service 220.
FIG. 2B illustrates data flow within a clustered virtualization environment implementing a VFS instance 202 in which stored items such as files and folders used by user VMs 101 are stored locally on the same host machines 200 as the user VMs 101 according to particular embodiments. As described above, one or more user VMs 101 and a Controller/Service VM 110 may run on each host machine 200 along with a hypervisor 130. As a user VM 101 processes I/O commands (e.g., a read or write operation), the I/O commands may be sent to the hypervisor 130 on the same server or host machine 200 as the user VM 101. For example, the hypervisor 130 may present to the user VMs 101 a VFS instance 202, receive an I/O command, and facilitate the performance of the I/O command by passing the command to a FSVM 170 that performs the operation specified by the command. The VFS 202 may facilitate I/O operations between a user VM 101 and a virtualized file system. The virtualized file system may appear to the user VM 101 as a namespace of mappable shared drives or mountable network file systems of files and directories. The namespace of the virtualized file system may be implemented using storage devices in the local storage 122, such as disks 204, onto which the shared drives or network file systems, files, and folders, or portions thereof, may be distributed as determined by the FSVMs 170. The VFS 202 may thus provide features disclosed herein, such as efficient use of the disks 204, high availability, scalability, and others. The implementation of these features may be transparent to the user VMs 101, 102. The FSVMs 170 may present the storage capacity of the disks 204 of the host machines 200 as an efficient, highly-available, and scalable namespace in which the user VMs 101, 102 may create and access shares, files, folders, and the like.
As an example, a network share may be presented to a user VM 101 as one or more discrete virtual disks, but each virtual disk may correspond to any part of one or more virtual or physical disks 204 within storage pool 160. Additionally or alternatively, the FSVMs 170 may present a VFS 202 either to the hypervisor 130 or to user VMs 101 of a host machine 200 to facilitate I/O operations. The FSVMs 170 may access the local storage 122 via Controller/Service VMs 110. As described above with reference to FIG. 1B, a Controller/Service VM 110 a may have the ability to perform I/O operations using local storage 122 a within the same host machine 200 a by connecting via the network 140 to cloud storage 126 or networked storage 128, or by connecting via the network 140 to local storage 122 b-c within another host machine 200 b-c (e.g., by connecting to another Controller/Service VM 110 b-c).
In particular embodiments, each user VM 101 may access one or more virtual disk images 206 stored on one or more disks 204 of the local storage 122, the cloud storage 126, and/or the networked storage 128. The virtual disk images 206 may contain data used by the user VMs 101, such as operating system images, application software, and user data, e.g., user home folders and user profile folders. For example, FIG. 2B illustrates three virtual machine images 206 a-c. The virtual machine image 206 a may be a file named UserVM101a.vmdisk (or the like) stored on disk 204 a of local storage 122 a of host machine 200 a. The virtual machine image 206 a may store the contents of the user VM 101 a's hard drive. The disk 204 a on which the virtual machine image 206 a is “local to” the user VM 101 a on host machine 200 a because the disk 204 a is in local storage 122 a of the host machine 200 a on which the user VM 101 a is located. Thus, the user VM 101 a may use local (intra-host machine) communication to access the virtual machine image 206 a more efficiently, e.g., with less latency and higher throughput, than would be the case if the virtual machine image 206 a were stored on disk 204 b of local storage 122 b of a different host machine 200 b, because inter-host machine communication across the network 140 would be used in the latter case. Local communication within a host machine 200 a is described in further detail with reference to FIG. 4C. Similarly, a virtual machine image 206 b, which may be a file named UserVM101b.vmdisk (or the like), is stored on disk 204 b of local storage 122 b of host machine 200 b, and the image 206 b is local to the user VM 101 b located on host machine 200 b. Thus, the user VM 101 a may access the virtual machine image 206 b more efficiently than the virtual machine 206 a on host machine 200 a, for example. In another example, the CVM 110 c may be located on the same host machine 200 c as the user VM 101 c that accesses a virtual machine image 206 c (UserVM101c.vmdisk) of the user VM 101 c, with the virtual machine image file 206 c being stored on a different host machine 200 b than the user VM 101 c and the CVM 110 c. In this example, communication between the user VM 101 c and the CVM 110 c may still be local, e.g., more efficient than communication between the user VM 101 c and a CVM 110 b on a different host machine 200 b, but communication between the CVM 110 c and the disk 204 b on which the virtual machine image 206 c is stored is via the network 140, as shown by the dashed lines between CVM 110 c and the network 140 and between the network 140 and local storage 122 b. The communication between CVM 110 c and the disk 204 b is not local, and thus may be less efficient than local communication such as may occur between the CVM 110 c and a disk 204 c in local storage 122 c of host machine 200 c. Further, a user VM 101 c on host machine 200 c may access data such as the virtual disk image 206 c stored on a remote (e.g., non-local) disk 204 b via network communication with a CVM 110 b located on the remote host machine 200 b. This case may occur if CVM 110 c is not present on host machine 200 c, e.g., because CVM 110 c has failed, or if the FSVM 170 c has been configured to communicate with local storage 122 b on host machine 200 b via the CVM 110 b on host machine 200 b, e.g., to reduce computational load on host machine 200 c.
In particular embodiments, since local communication is expected to be more efficient than remote communication, the FSVMs 170 may store storage items, such as files or folders, e.g., the virtual disk images 206, on local storage 122 of the host machine 200 on which the user VM 101 that is expected to access the files is located. A user VM 101 may be expected to access particular storage items if, for example, the storage items are associated with the user VM 101, such as by configuration information. For example, the virtual disk image 206 a may be associated with the user VM 101 a by configuration information of the user VM 101 a. Storage items may also be associated with a user VM 101 via the identity of a user of the user VM 101. For example, files and folders owned by the same user ID as the user who is logged into the user VM 101 a may be associated with the user VM 101 a. If the storage items expected to be accessed by a user VM 101 a are not stored on the same host machine 200 a as the user VM 101 a, e.g., because of insufficient available storage capacity in local storage 122 a of the host machine 200 a, or because the storage items are expected to be accessed to a greater degree (e.g., more frequently or by more users) by a user VM 101 b on a different host machine 200 b, then the user VM 101 a may still communicate with a local CVM 110 a to access the storage items located on the remote host machine 200 b, and the local CVM 110 a may communicate with local storage 122 b on the remote host machine 200 b to access the storage items located on the remote host machine 200 b. If the user VM 101 a on a host machine 200 a does not or cannot use a local CVM 110 a to access the storage items located on the remote host machine 200 b, e.g., because the local CVM 110 a has crashed or the user VM 101 a has been configured to use a remote CVM 110 b, then communication between the user VM 101 a and local storage 122 b on which the storage items are stored may be via a remote CVM 110 b using the network 140, and the remote CVM 110 b may access local storage 122 b using local communication on host machine 200 b. As another example, a user VM 101 a on a host machine 200 a may access storage items located on a disk 204 c of local storage 122 c on another host machine 200 c via a CVM 110 b on an intermediary host machine 200 b using network communication between the host machines 200 a and 200 b and between the host machines 200 b and 200 c.
In particular embodiments, VFS 202 may provide access controls for accessing storage items in a distributed file share of storage items. The distributed file share of storage items includes files, folders, etc., that are managed across the FSVMs 170 a-c of the VFS 202. The access controls may be stored for each of the storage item may be specified in access control lists (ACLs) including permissions information for individual users and groups. In particular embodiments, such metadata may be stored in a database and then cached when a storage access operation referencing the storage item is received. An ACL may comprise a list of access control entries (ACE). Each ACE in an ACL identifies a user and specifies the access rights allowed, denied, or audited for that user. For example, in a Windows environment, each ACL may be on average 4 KB in size, but may grow up to 64 KB in size.
By way of example and not limitation, as described in the specification for MICROSOFT SMB protocol versions 2 and 3 (v20160714, released Jul. 14, 2016), permissions information for accessing a file, pipe, or printer may be stored in the form of a 4-byte bitmask (a small set of Boolean values):


File_Pipe_Printer_Access_Mask

0	1	2	3	4	5	6	7	8	9	10	1	2	3	4	5	6	7	8	9	20	1	2	3	4	5	6	7	8	9	30	1

Such a bitmask may enable specification of the following types of permissions:


FILE_READ_DATA	Indicates the right to read data from the file or named pipe
0x00000001
FILE_WRITE_DATA	Indicates the right to write data into the file or named pipe
0x00000002	beyond the end of the file
FILE_APPEND_DATA	Indicates the right to append data into the file or named pipe
0x00000004
FILE_READ_EA	Indicates the right to read the extended attributes of the
0x00000008	file or named pipe
FILE_WRITE_EA	Indicates the right to write or change the extended
0x00000010	attributes of the file or named pipe
FILE_DELETE_CHILD	Indicates the right to delete entries within a directory
0x00000040
FILE_EXECUTE	Indicates the right to execute the file
0x00000020
FILE_READ_ATTRIBUTES	Indicates the right to read the attributes of the file
0x00000080
FILE_WRITE_ATTRIBUTES	Indicates the right to change attributes of the file
0x00000100
DELETE	Indicates the right to delete the file
0x00010000
READ_CONTROL	Indicates the right to read the security descriptor for the
0x00020000	file or named pipe
WRITE_DAC	Indicates the right to change the discretionary access
0x00040000	control list (DACL) in the security descriptor for the file
	or named pipe
WRITE_OWNER	Indicates the right to change the owner in the security
0x00080000	descriptor for the file or named pipe
SYNCHRONIZE	SMB2 clients set this flag to any value.
0x00100000	SMB2 servers should ignore this flag.
ACCESS_SYSTEM_SECURITY	Indicates the right to read or change the system access
0x01000000	control list (SACL) in the security descriptor for the file or
	named pipe
MAXIMUM_ALLOWED	Indicates that the client is requesting an open to the file
0x02000000	with the highest level of access the client has on this file
GENERIC_ALL	Indicates a request for all the access flags that are
0x10000000	previously listed except MAXIMUM_ALLOWED and
	ACCESS_SYSTEM_SECURITY
GENERIC_EXECUTE	Indicates a request for the following combination of
0x20000000	access flags listed above: FILE_READ_ATTRIBUTES \|
	FILE_EXECUTE \| SYNCHRONIZE \|
	READ_CONTROL.
GENERIC_WRITE	Indicates a request for the following combination of
0x40000000	access flags listed above: FILE_WRITE_DATA \|
	FILE_APPEND_DATA \| FILE_WRITE_ATTRIBUTES \|
	FILE_WRITE_EA \| SYNCHRONIZE \|
	READ_CONTROL.
GENERIC_READ	Indicates a request for the following combination of
0x80000000	access flags listed above: FILE_READ_DATA \|
	FILE_READ_ATTRIBUTES \| FILE_READ_EA \|
	SYNCHRONIZE \| READ_CONTROL.

Further by way of example and not limitation, as described in the specification for MICROSOFT SMB protocol versions 2 and 3 (v20160714, released Jul. 14, 2016), permissions information for accessing a directory may be stored in the form of a 4-byte bitmask:


Directory_Access_Mask

0	1	2	3	4	5	6	7	8	9	10	1	2	3	4	5	6	7	8	9	20	1	2	3	4	5	6	7	8	9	30	1

Such a bitmask may enable specification of the following types of permissions:


FILE_LIST_DIRECTORY	Indicates the right to enumerate contents of the directory
0x00000001
FILE_ADD_FILE	Indicates the right to create a file under the directory
0x00000002
FILE_ADD_SUBDIRECTORY	Indicates the right to add a sub-directory under the
0x00000004	directory
FILE_READ_EA	Indicates the right to read the extended attributes of the
0x00000008	directory
FILE_WRITE_EA	Indicates the right to write or change the extended
0x00000010	attributes of the directory
FILE_TRAVERSE	Indicates the right to traverse this directory if the server
0x00000020	enforces traversal checking
FILE_DELETE_CHILD	Indicates the right to delete the files and directories within
0x00000040	this directory
FILE_READ_ATTRIBUTES	Indicates the right to read the attributes of the directory
0x00000080
FILE_WRITE_ATTRIBUTES	Indicates the right to change attributes of the directory
0x00000100
DELETE	Indicates the right to delete the directory
0x00010000
READ_CONTROL	Indicates the right to read the security descriptor for the
0x00020000	directory
WRITE_DAC	Indicates the right to change the discretionary access
0x00040000	control list (DACL) in the security descriptor for the
	directory
WRITE_OWNER	Indicates the right to change the owner in the security
0x00080000	descriptor for the directory
SYNCHRONIZE	SMB2 clients set this flag to any value.
0x00100000	SMB2 servers should ignore this flag.
ACCESS_SYSTEM_SECURITY	Indicates the right to read or change the system access
0x01000000	control list (SACL) in the security descriptor for the
	directory
MAXIMUM_ALLOWED	Indicates that the client is requesting an open to the
0x02000000	directory with the highest level of access the client has on
	this directory
GENERIC_ALL	Indicates a request for all the access flags that are
0x10000000	previously listed except MAXIMUM_ALLOWED and
	ACCESS_SYSTEM_SECURITY
GENERIC_EXECUTE	Indicates a request for the following combination of
0x20000000	access flags listed above: FILE_READ_ATTRIBUTES \|
	FILE_TRAVERSE \| SYNCHRONIZE \|
	READ_CONTROL.
GENERIC_WRITE	Indicates a request for the following combination of
0x40000000	access flags listed above: FILE_ADD_FILE \|
	FILE_ADD_SUBDIRECTORY \|
	FILE_WRITE_ATTRIBUTES \| FILE_WRITE_EA \|
	SYNCHRONIZE \| READ_CONTROL.
GENERIC_READ	Indicates a request for the following combination of
0x80000000	access flags listed above: FILE_LIST_DIRECTORY \|
	FILE_READ_ATTRIBUTES \| FILE_READ_EA \|
	SYNCHRONIZE \| READ_CONTROL.

When the ACL for a storage item is large in size (e.g., as may be the case where there are many ACEs listed for different users/groups and/or many different types of permissions that may be specified), retrieving the permissions information may be an expensive operation, particularly when enumerating (e.g., listing) the contents of a directory or a distributed file share. Given that the permissions information must be accessed with each and every storage access operation, the cost may become considerable. In particular, a simple operation such as enumerating the contents of a directory or a distributed file share presents a particular challenge, since permissions must be checked for every storage item (e.g., files or directories) within the directory or distributed file share, in order to determine whether the user should be allowed to view the names of all the storage items in the directory or distributed file share.
When receiving a request to enumerate the contents of a directory or a distributed file share, particular embodiments may allocate a sufficiently large buffer to contain metadata (including permissions information) for a large set of the storage items (e.g., 5K-10K directories) in the directory or distributed file share. Metadata may be fetched for this first set of storage items in a single request (e.g., query to a database), then some or all of the metadata may be stored in the buffer in a pre-defined format, such as, by way of example and not limitation: |dir_1 name length|dir_1 name|dir_1 stat size|dir_1 stat|dir_1 ACL size|dir_1 ACL|dir_2 name length|dir_2 name|dir_2 stat size|dir_2 stat|dir_2 ACL size|dir_2 ACL| . . . |dir_n name length|dir_n name|dir_n stat size|dir_n stat|dir_n ACL size|dir_n ACL|. Particular embodiments may iterate over entries in this buffer, checking the permissions information of each storage item in the set, and building a response to the request. For each storage item in the set, particular embodiments may compare the requesting user's access controls with the ACL to determine if the user has permissions to view the storage item—if the user has permissions to view the storage item, then the storage item is included in the response, otherwise, particular embodiments may skip to evaluate the ACL for the next storage item in the set. Once permissions information for all the storage items in the buffer have been evaluated, particular embodiments may fetch the next set of storage items in the directory or distributed file share, and repeat the above-described procedure until all permissions information for all the storage items in the directory or distributed file share have been evaluated.
Particular embodiments may discover a local and writable domain controller in a multi-site and multi-domain controller (e.g., ACTIVE DIRECTORY) distributed across geographical locations, rather than requiring user input to select a domain controller. Particular embodiments may query (e.g., an LDAP UDP Netlogon ping) the given domain in order to find the local site name for the network. From the local site name, particular embodiments may find all the domain controllers in the local site (e.g., by doing an SRV DNS lookup). Particular embodiments may successively poll (e.g., an LDAP UDP Netlogon ping) each of the domain controllers in the local site until a response is received from a writable domain controller. This writable domain controller may be used for all domain controller communication going forward.
Particular embodiments may enable user VMs to send referral requests to FSVMs 170 when attempting to access a particular directory. The referral request may specify a path to the directory. When resolving the path to determine the exact location of the directory, particular embodiments may need to locate which host machine is storing the directory (and the data/files in the directory). Metadata for attributes of storage items that is captured in the snapshot of the consistency group for the FSVMs may be stored in a database. Such metadata may include, by way of example and not limitation: file type, permissions, owner, group, file size, file access time, file modification time, file deletion time, number of (soft/hard) links to the file/directory, extended attributes, or an access control list (ACL).
In conventional embodiments, for a directory, the owner designation metadata may refer to the host machine on which the directory (and the data/files in the directory) is physically stored on disk. This direct mapping scheme may be burdensome to maintain in a virtualized file system, where volume groups (and their vdisks) containing the directory are (1) managed independently of the CVMs to which they are attached and (2) may be shifted between host machines in order to rebalance load on FSVMs and/or CVMs. In such embodiments, every time that a volume group is moved, the owner designation metadata must be updated in all directories in the volume group.
In order to eliminate this overhead, particular embodiments may establish a two-level mapping scheme, whereby the owner designation metadata for each directory indicates the volume group with which the directory is associated (rather than specifying any host machine), and then a mapping between the volume group and the host machine associated with the volume group is stored separately (e.g., in a separate lookup table). When a referral request is received, particular embodiments may resolve the path by first extracting the owner designation metadata, then performing a lookup to identify the host machine on which the directory's data resides. Whenever a volume group is moved from one host machine to another, only that mapping (e.g., in the lookup table) need be updated.
FIG. 3A illustrates an example hierarchical structure 300 of a VFS instance in a cluster according to particular embodiments. A Cluster 302 contains two VFS instances, FS1 304 and FS2 306. Each VFS instance may be identified by a name such as “\\instance”, e.g., “\\FS1” for WINDOWS file systems, or a name such as “instance”, e.g., “FS1” for UNIX-type file systems. The VFS instance FS1 304 contains shares, including Share-1 308 and Share-2 310. Shares may have names such as “Users” for a share that stores user home directories, or the like. Each share may have a path name such as \\FS1\Share-1 or \\FS1\Users. As an example and not by way of limitation, a share may correspond to a disk partition or a pool of file system blocks on WINDOWS and UNIX-type file systems. As another example and not by way of limitation, a share may correspond to a folder or directory on a VFS instance 304. Shares may appear in the file system instance 202 as folders or directories to users of user VMs 101 a. Share-1 308 includes two folders, Folder-1 312, and Folder-2 314, and may also include one or more files (e.g., files not in folders). Each folder 312, 314 may include one or more files 318. Share-2 310 includes a folder Folder-3 316, which includes a file File-2 320. Each folder has a folder name such as “Folder-1”, “Users”, or “Sam” and a path name such as “\\FS1\Share-1\Folder-1” (WINDOWS) or “share-1:/fs1/Users/Sam” (UNIX). Similarly, each file has a file name such as “File-1” or “Forecast.xls” and a path name such as “\\FS1\Share-1 \Folder-1 \File-1” or “share-1:/fs1/Users/Sam/Forecast.xls”.
FIG. 3B illustrates two example host machines 200 a and 200 b, each providing file storage services for portions of two VFS instances FS1 and FS2 according to particular embodiments. The first host machine, Host-1 200 a, includes two user VMs 101 a, 102 a, a Hypervisor 130 a, a FSVM named FileServer-VM-1 (abbreviated FSVM-1) 170 a, a Controller/Service VM named CVM-1 110 a, and local storage 122 a. Host-1's FileServer-VM-1 170 a has an IP (Internet Protocol) network address of 10.1.1.1, which is an address of a network interface on Host-1 200 a. Host-1 has a hostname ip-addr1, which may correspond to Host-1's IP address 10.1.1.1. The second host machine, Host-2 200 b, includes two user VMs 101 b, 102 b, a Hypervisor 130 b, a File Server VM named FileServer-VM-2 (abbreviated FSVM-2) 170 b, a Controller/Service VM named CVM-2 110 b, and local storage 122 b. Host-2's FileServer-VM-1 170 b has an IP network address of 10.1.1.2, which is an address of a network interface on Host-2 200 b.
In particular embodiments, file systems FileSystem-1A 364 a and FileSystem-2A 365 a implement the structure of files and folders for portions of the FS1 and FS2 file server instances, respectively, that are located on (e.g., served by) FileServer-VM-1 170 a on Host-1 200 a. Other file systems on other host machines may implement other portions of the FS1 and FS2 file server instances. The file systems 364 a and 365 a may implement the structure of at least a portion of a file server instance by translating file system operations, such as opening a file, writing data to or reading data from the file, deleting a file, and so on, to disk I/O operations such as seeking to a portion of the disk, reading or writing an index of file information, writing data to or reading data from blocks of the disk, allocating or de-allocating the blocks, and so on. The file systems 364 a, 365 a may thus store their file system data, including the structure of the folder and file hierarchy, the names of the storage items (e.g., folders and files), and the contents of the storage items on one or more storage devices, such as local storage 122 a. The particular storage device or devices on which the file system data for each file system are stored may be specified by an associated file system pool (e.g., 366 a-c and 367 a-c). For example, the storage device(s) on which data for FileSystem-1A 364 a and FileSystem-2A, 365 a are stored may be specified by respective file system pools FS1-Pool-1 366 a and FS2-Pool-2 367 a. The storage devices for the pool 366 a may be selected from volume groups provided by CVM-1 110 a, such as volume group VG1 368 a and volume group VG2 369 a. Each volume group 368 a, 369 a may include a group of one or more available storage devices that are present in local storage 122 a associated with (e.g., by iSCSI communication) the CVM-1 110 a. The CVM-1 110 a may be associated with a local storage 122 a on the same host machine 200 a as the CVM-1 110 a, or with a local storage 122 b on a different host machine 200 b. The CVM-1 110 a may also be associated with other types of storage, such as cloud storage 126, networked storage 128 or the like. Although the examples described herein include particular host machines, virtual machines, file servers, file server instances, file server pools, CVMs, volume groups, and associations there between, any number of host machines, virtual machines, file servers, file server instances, file server pools, CVMs, volume groups, and any associations there between are possible and contemplated.
In particular embodiments, the file system pool 366 a may associate any storage device in one of the volume groups 368 a, 369 a of storage devices that are available in local storage 122 a with the file system FileSystem-1A 364 a. For example, the file system pool FS1-Pool-1 366 a may specify that a disk device named hd1 in the volume group VG1 368 a of local storage 122 a is a storage device for FileSystem-1A 364 a for file server FS1 on FSVM-1 170 a. A file system pool FS2-Pool-2 367 a may specify a storage device FileSystem-2A 365 a for file server FS2 on FSVM-1 170 a. The storage device for FileSystem-2A 365 a may be, e.g., the disk device hd1, or a different device in one of the volume groups 368 a, 369 a, such as a disk device named hd2 in volume group VG2 369 a. Each of the file systems FileSystem-1A 364 a, FileSystem-2A 365 a may be, e.g., an instance of the NTFS file system used by the WINDOWS operating system, of the UFS Unix file system, or the like. The term “file system” may also be used herein to refer to an instance of a type of file system, e.g., a particular structure of folders and files with particular names and content.
In one example, referring to FIG. 3A, an FS1 hierarchy rooted at File Server FS1 304 may be located on FileServer-VM-1 170 a and stored in file system instance FileSystem-1A 364 a. That is, the file system instance FileSystem-1A 364 a may store the names of the shares and storage items (such as folders and files), as well as the contents of the storage items, shown in the hierarchy at and below File Server FS1 304. A portion of the FS1 hierarchy shown in FIG. 3A, such the portion rooted at Folder-2 314, may be located on FileServer-VM-2-170 b on Host-2 200 b instead of FileServer-VM-1-170 a, in which case the file system instance FileSystem-1B 364 b may store the portion of the FS1 hierarchy rooted at Folder-2 314, including Folder-3 314, Folder-4 322 and File-3 324. Similarly, an FS2 hierarchy rooted at File Server FS2 306 in FIG. 3A may be located on FileServer-VM-1 170 a and stored in file system instance FileSystem-2A 365 a. The FS2 hierarchy may be split into multiple portions (not shown), such that one portion is located on FileServer-VM-1 170 a on Host-1 200 a, and another portion is located on FileServer-VM-2 170 b on Host-2 200 b and stored in file system instance FileSystem-2B 365 c.
In particular embodiments, FileServer-VM-1 (abbreviated FSVM-1) 170 a on Host-1 200 a is a leader for a portion of file server instance FS1 and a portion of FS2, and is a backup for another portion of FS1 and another portion of FS2. The portion of FS1 for which FileServer-VM-1 170 a is a leader corresponds to a storage pool labeled FS1-Pool-1 366 a. FileServer-VM-1 is also a leader for FS2-Pool-2 367 a, and is a backup (e.g., is prepared to become a leader upon request, such as in response to a failure of another FSVM) for FS1-Pool-3 366 b and FS2-Pool-4 367 b on Host-2. In particular embodiments, FileServer-VM-2 (abbreviated FSVM-2) 170 b is a leader for a portion of file server instance FS1 and a portion of FS2, and is a backup for another portion of FS1 and another portion of FS2. The portion of FS1 for which FSVM-2 170 b is a leader corresponds to a storage pool labeled FS1-Pool-3 366 b. FSVM-2 170 b is also a leader for FS2-Pool-4 367 b, and is a backup for FS1-Pool-1 366 a and FS2-Pool-2 367 a on Host-1.
In particular embodiments, the file server instances FS1, FS2 provided by the FSVMs 170 a and 170 b may be accessed by user VMs 101 a and 101 b via a network file system protocol such as SMB, CIFS, NFS, or the like. Each FSVM 170 a and 170 b may provide what appears to client applications on user VMs 101 a and 101 b to be a single file system instance, e.g., a single namespace of shares, files and folders, for each file server instance 202. However, the shares, files, and folders in a file server instance such as FS1 may actually be distributed across multiple FSVMs 170 a and 170 b. For example, different folders in the same file server instance may be associated with different corresponding FSVMs 170 a and 170 b and CVMs 110 a and 110 b on different host machines 200 a and 200 b.
The example file server instance FS1 304 shown in FIG. 3A has two shares, Share-1 308 and Share-2 310. Share-1 308 may be located on FSVM-1 170 a, CVM-1 110 a, and local storage 122 a. Network file system protocol requests from user VMs 101 and 102 to read or write data on file server instance FS1 304 and any share, folder, or file in the instance may be sent to FSVM-1 170 a. FSVM-1 170 a may determine whether the requested data, e.g., the share, folder, file, or a portion thereof, referenced in the request, is located on FSVM-1, and FSVM-1 is a leader for the requested data. If not, FSVM-1 may respond to the requesting User-VM with an indication that the requested data is not covered by (e.g., is not located on or served by) FSVM-1. Otherwise, the requested data is covered by (e.g., is located on or served by) FSVM-1, so FSVM-1 may send iSCSI protocol requests to a CVM that is associated with the requested data. Note that the CVM associated with the requested data may be the CVM-1 110 a on the same host machine 200 a as the FSVM-1, or a different CVM on a different host machine 200 b, depending on the configuration of the VFS 202. In this example, the requested Share-1 is located on FSVM-1, so FSVM-1 processes the request. To provide for path availability, multipath I/O (MPIO) may be used for communication with the FSVM, e.g., for communication between FSVM-1 and CVM-1. The active path may be set to the CVM that is local to the FSVM (e.g., on the same host machine) by default. The active path may be set to a remote CVM instead of the local CVM, e.g., when a failover occurs.
Continuing with the data request example, the associated CVM is CVM 110 a, which may in turn access the storage device associated with the requested data as specified in the request, e.g., to write specified data to the storage device or read requested data from a specified location on the storage device. In this example, the associated storage device is in local storage 122 a, and may be an HDD or SSD. CVM-1 110 a may access the HDD or SSD via an appropriate protocol, e.g., iSCSI, SCSI, SATA, or the like. CVM 110 a may send the results of accessing local storage 122 a, e.g., data that has been read, or the status of a data write operation, to CVM 110 a via, e.g., SATA, which may in turn send the results to FSVM-1 170 a via, e.g., iSCSI. FSVM-1 170 a may then send the results to user VM 101 a via SMB through the Hypervisor 130 a.
Share-2 310 may be located on FSVM-2 170 b, on Host-2. Network file service protocol requests from user VMs 101 a and 101 b to read or write data on Share-2 may be directed to FSVM-2 170 b on Host-2 by other FSVMs 170 a. Alternatively, user VMs 101 a and 101 b may send such requests directly to FSVM-2 170 b on Host-2, which may process the requests using CVM-2 110 b and local storage 122 b on Host-2 as described above for FSVM-1 170 a on Host-1.
A file server instance 202 such as FS1 304 in FIG. 3A may appear as a single file system instance (e.g., a single namespace of folders and files that are accessible by their names or pathnames without regard for their physical locations), even though portions of the file system are stored on different host machines 200 a-c. Since each FSVM 170 may provide a portion of a file server instance 202, each FSVM 170 may have one or more “local” file systems 364 a, 365 a that provide the portion of the file server instance 202 (e.g., the portion of the namespace of files and folders) associated with the FSVM 170.
FIG. 3C illustrates example interactions between a client 330 and host machines 200 a and 200 c on which different portions of a VFS instance are stored according to particular embodiments. A client 330, e.g., an application program executing in one of the user VMs 101 and 102 on the host machines 200 a-c of FIGS. 2A-2B (e.g. user VM 101 b on host machine 200 b) requests access to a folder \\FS1.domain.name\Share-1\Folder-3. The request may be in response to an attempt to map \\FS1.domain.name\Share-1 to a network drive in the operating system executing in the user VM 101 c followed by an attempt to access the contents of Share-1 or to access the contents of Folder-3, such as listing the files in Folder-3.
FIG. 3C shows interactions that occur between the client 330, FSVMs 170 a and 170 b on host machines 200 a and 200 b, and a name server 332 when a storage item is mapped or otherwise accessed. The name server 332 may be provided by a server computer system, such as one or more of the host machines 200, or a server computer system separate from the host machines 200. In one example, the name server 332 may be provided by an ACTIVE DIRECTORY service executing on one or more computer systems and accessible via the network 140. The interactions are shown as arrows that represent communications, e.g., messages sent via the network 140. Note that the client 330 may be executing in a user VM 101, which may be co-located with one of the FSVMs 170 a and 170 b. In such a co-located case, the arrows between the client 330 and the host machine 200 on which the FSVM 170 is located may represent communication within the host machine 200, and such intra-host machine communication may be performed using a mechanism different from communication over the network 140, e.g., shared memory or inter process communication.
In particular embodiments, when the client 330 requests access to Folder-3, a VFS client component executing in the user VM 101 b may use a distributed file system protocol such as MICROSOFT DFS, or the like, to send the storage access request to one or more of the FSVMs 170 a-c of FIGS. 2A-2B. To access the requested file or folder, the client determines the location of the requested file or folder, e.g., the identity and/or network address of the FSVM 170 on which the file or folder is located. The client may query a domain cache of FSVM 170 a-c network addresses that the client has previously identified (e.g., looked up). If the domain cache contains the network address of an FSVM 170 associated with the requested folder name \\FS1.domain.name\Share-1\Folder-3, then the client retrieves the associated network address from the domain cache and sends the access request to the network address, starting at step 393 as described below.
In particular embodiments, at step 381, the client may send a request for a list of addresses of FSVMs 170 a-170 c to a name server 332. The name server 332 may be, e.g., a DNS server or other type of server, such as a MICROSOFT domain controller (not shown), that has a database of FSVM addresses. At step 382, the name server 332 may send a reply that contains a list of FSVM 170 network addresses, e.g., ip-addr1, ip-addr2, and ip-addr3, which correspond to the FSVMs 170 a-c in this example. At step 383, the client 330 may send an access request to one of the network addresses, e.g., the first network address in the list (ip-addr1 in this example), requesting the contents of Folder-3 of Share-1. By selecting the first network address in the list, the particular FSVM 170 to which the access request is sent may be varied, e.g., in a round-robin manner by enabling round-robin DNS (or the like) on the name server 332. The access request may be, e.g., an SMB connect request, an NFS open request, and/or appropriate request(s) to traverse the hierarchy of Share-1 to reach the desired folder or file, e.g., Folder-3 in this example.
At step 384, FileServer-VM-1 170 a may process the request received at step 383 by searching a mapping or lookup table, such as a sharding map 360 a, for the desired folder or file. The map 360 maps stored objects, such as shares, folders, or files, to their corresponding locations, e.g., the names or addresses of FSVMs 170. The map 360 may have the same contents on each host machine 200, with the contents on different host machines being synchronized using a distributed data store as described below. For example, the map 360 a may contain entries that map Share-1 and Folder-1 to the File Server FSVM-1 170 a, and Folder-3 to the File Server FSVM-3 170 c. An example map 360 is shown in Table 1 below.

TABLE 1

	Stored Object	Location

	Folder-1	FSVM-1
	Folder-2	FSVM-1
	File-1	FSVM-1
	Folder-3	FSVM-3
	File-2	FSVM-3

In particular embodiments, the map 360 may be accessible on each of the host machines 200. As described with reference to FIGS. 2A-2B, the maps 360 a and 360 c may be copies of a distributed data structure that are maintained and accessed at each FSVM 170 a-c using a distributed data access coordinator 370 a and 370 c. The distributed data access coordinator 370 a and 370 c may be implemented based on distributed locks or other storage item access operations. Alternatively, the distributed data access coordinator 370 a and 370 c may be implemented by maintaining a master copy of the maps 360 a and 360 c at a leader node such as the host machine 200 c, and using distributed locks to access the master copy from each FSVM 170 a and 170 b. The distributed data access coordinator 370 a and 370 c may be implemented using distributed locking, leader election, or related features provided by a centralized coordination service for maintaining configuration information, naming, providing distributed synchronization, and/or providing group services (e.g., APACHE ZOOKEEPER or other distributed coordination software). Since the map 360 a indicates that Folder-3 is located at FSVM-3 170 c on Host-3 200 c, the lookup operation at step 384 determines that Folder-3 is not located at FSVM-1 on Host-1 200 a. Thus, at step 385 the FSVM-1 170 a sends a response, e.g., a “Not Covered” DFS response, to the client 330 indicating that the requested folder is not located at FSVM-1. At step 386, the client 330 sends a request to FSVM-1 for a referral to the FSVM on which Folder-3 is located. FSVM-1 uses the map 360 a to determine that Folder-3 is located at FSVM-3 on Host-3 200 c, and at step 387 returns a response, e.g., a “Redirect” DFS response, redirecting the client 330 to FSVM-3. The client 330 may then determine the network address for FSVM-3, which is ip-addr3 (e.g., a host name “ip-addr3.domain.name” or an IP address, 10.1.1.3). The client 330 may determine the network address for FSVM-3 by searching a cache stored in memory of the client 330, which may contain a mapping from FSVM-3 to ip-addr3 cached in a previous operation. If the cache does not contain a network address for FSVM-3, then at step 388 the client 330 may send a request to the name server 332 to resolve the name FSVM-3. The name server may respond with the resolved address, ip-addr3, at step 389. The client 330 may then store the association between FSVM-3 and ip-addr3 in the client's cache.
In particular embodiments, failure of FSVMs 170 may be detected using the centralized coordination service. For example, using the centralized coordination service, each FSVM 170 a may create a lock on the host machine 200 a on which the FSVM 170 a is located using ephemeral nodes of the centralized coordination service (which are different from host machines 200 but may correspond to host machines 200). Other FSVMs 170 b and 170 c may volunteer for leadership of resources of remote FSVMs 170 on other host machines 200, e.g., by requesting a lock on the other host machines 200. The locks requested by the other nodes are not granted unless communication to the leader host machine 200 c is lost, in which case the centralized coordination service deletes the ephemeral node and grants the lock to one of the volunteer host machines 200 a and 200 b, which becomes the new leader. For example, the volunteer host machines 200 a and 200 b may be ordered by the time at which the centralized coordination service received their requests, and the lock may be granted to the first host machine 200 on the ordered list. The first host machine 200 (e.g., host machine 200 b) on the list may thus be selected as the new leader. The FSVM 170 b on the new leader has ownership of the resources that were associated with the failed leader FSVM 170 a until the failed leader FSVM 170 c is restored, at which point the restored FSVM 170 a may reclaim the local resources of the host machine 200 c on which it is located.
At step 390, the client 330 may send an access request to FSVM-3 170 c at ip-addr3 on Host-3 200 c requesting the contents of Folder-3 of Share-1. At step 391, FSVM-3 170 c queries FSVM-3's copy of the map 360 using FSVM-3's instance of the distributed data access coordinator 370 c. The map 360 indicates that Folder-3 is located on FSVM-3, so at step 392 FSVM-3 accesses the file system 364 c to retrieve information about Folder-3 316 and its contents (e.g., a list of files in the folder, which includes File-2 320) that are stored on the local storage 122 c. FSVM-3 may access local storage 122 c via CVM-3 110 c, which provides access to local storage 122 c via a volume group 368 c that contains one or more volumes stored on one or more storage devices in local storage 122 c. At step 393, FSVM-3 may then send the information about Folder-3 and its contents to the client 330. Optionally, FSVM-3 may retrieve the contents of File-2 and send them to the client 330, or the client 330 may send a subsequent request to retrieve File-2 as needed.
FIG. 3D illustrates an example virtualized file server having a failover capability according to particular embodiments. To provide high availability, e.g., so that the file server continues to operate after failure of components such as a CVM, FSVM, or both, as may occur if a host machine fails, components on other host machines may take over the functions of failed components. When a CVM fails, a CVM on another host machine may take over input/output operations for the failed CVM. Further, when an FSVM fails, an FSVM on another host machine may take over the network address and CVM or volume group that were being used by the failed FSVM. If both an FSVM and an associated CVM on a host machine fail, as may occur when the host machine fails, then the FSVM and CVM on another host machine may take over for the failed FSVM and CVM. When the failed FSVM and/or CVM are restored and operational, the restored FSVM and/or CVM may take over the operations that were being performed by the other FSVM and/or CVM. In FIG. 3D, FSVM-1 170 a communicates with CVM-1 110 a to use the data storage in volume groups VG1 368 a and VG2 369 a. For example, FSVM-1 is using disks in VG1 and VG2, which are iSCSI targets. FSVM-1 has iSCSI initiators that communicate with the VG1 and VG2 targets using MPIO (e.g., DM-MPIO on the LINUX operating system). FSVM-1 may access the volume groups VG1 and VG2 via in-guest iSCSI. Thus, any FSVM may connect to any iSCSI target if an FSVM failure occurs.
In particular embodiments, during failure-free operation, there are active iSCSI paths between FSVM-1 and CVM-1, as shown in FIG. 3D by the dashed lines from the FSVM-1 file systems for FS1 364 a and FS2 365 a to CVM-1's volume group VG1 368 a and VG2 369 a, respectively. Further, during failure-free operation there are inactive failover (e.g., standby) paths between FSVM-1 and CVM-3 110 c, which is located on Host-3. The failover paths may be, e.g., paths that are ready to be activated in response to the local CVM CVM-1 becoming unavailable. There may be additional failover paths that are not shown in FIG. 3D. For example, there may be failover paths between FSVM-1 and a CVM on another host machine, such as CVM-2 110 b on Host-2 200 b. The local CVM CVM-1 110 a may become unavailable if, for example, CVM-1 crashes, or the host machine on which the CVM-1 is located crashes, loses power, loses network communication between FSVM-1 170 a and CVM-1 110 a. As an example and not by way of limitation, the failover paths do not perform I/O operations during failure-free operation. Optionally, metadata associated with a failed CVM 110 a, e.g., metadata related to volume groups 368 a, 369 a associated with the failed CVM 110 a, may be transferred to an operational CVM, e.g., CVM 110 c, so that the specific configuration and/or state of the failed CVM 110 a may be re-created on the operational CVM 110 c.
FIG. 3E illustrates an example virtualized file server that has recovered from a failure of Controller/Service VM CVM-1 110 a by switching to an alternate Controller/Service VM CVM-3 110 c according to particular embodiments. When CVM-1 110 a fails or otherwise becomes unavailable, then the FSVM associated with CVM-1, FSVM-1 170 a, may detect a PATH DOWN status on one or both of the iSCSI targets for the volume groups VG1 368 a and VG2 369 a, and initiate failover to a remote CVM that can provide access to those volume groups VG1 and VG2. For example, when CVM-1 110 a fails, the iSCSI MPIO may activate failover (e.g., standby) paths to the remote iSCSI target volume group(s) associated with the remote CVM-3 110 c on Host-3 200 c. CVM-3 provides access to volume groups VG1 and VG2 as VG1 368 c and VG2 369 c, which are on storage device(s) of local storage 122 c. The activated failover path may take over I/O operations from failed CVM-1 110 a. Optionally, metadata associated with the failed CVM-1 110 a, e.g., metadata related to volume groups 368 a, 369 a, may be transferred to CVM-3 so that the specific configuration and/or state of CVM-1 may be re-created on CVM-3. When the failed CVM-1 again becomes available, e.g., after it has been re-started and has resumed operation, the path between FSVM-1 and CVM-1 may reactivated or marked as the active path, so that local I/O between CVM-1 and FSVM-1 may resume, and the path between CVM-3 and FSVM-1 may again become a failover (e.g., standby) path.
FIG. 3F illustrates an example virtualized file server that has recovered from failure of a FSVM by electing a new leader FSVM according to particular embodiments. When an FSVM-2 170 b fails, e.g., because it has been brought down for maintenance, has crashed, the host machine on which it was executing has been powered off or crashed, network communication between the FSVM and other FSVMs has become inoperative, or other causes, then the CVM that was being used by the failed FSVM, the CVM's associated volume group(s), and the network address of the host machine on which the failed FSVM was executing may be taken over by another FSVM to provide continued availability of the file services that were being provided by the failed FSVM. In the example shown in FIG. 3F, FSVM-2 170 b on Host-2 200 b has failed. One or more other FSVMs, e.g., FSVM-1 170 a or FSVM-3 170 c, or other components located on one or more other host machines, may detect the failure of FSVM-2, e.g., by detecting a communication timeout or lack of response to a periodic status check message. When FSVM-2's failure is detected, an election may be held, e.g., using a distributed leader election process such as that provided by the centralized coordination service. The host machine that wins the election may become the new leader for the file system pools 366 b, 367 b for which the failed FSVM-2 was the leader. In this example, FSVM-1 170 a wins the election and becomes the new leader for the pools 366 b, 367 b. FSVM-1 170 a thus attaches to CVM-2 110 b by creating file system 364 b, 365 c instances for the file server instances FS1 and FS2 using FS1-Pool-3 366 b and FS2-Pool-4 367 b, respectively. In this way, FSVM-1 takes over the file systems and pools for CVM-2's volume groups, e.g., volume groups VG1 366 b and VG2 367 b of local storage 122 b. Further, FSVM-1 takes over the IP address associated with FSVM-2, 10.1.1.2, so that storage access requests sent to FSVM-2 are received and processed by FSVM-1. Optionally, metadata used by FSVM-1, e.g., metadata associated with the file systems, may be transferred to FSVM-3 so that the specific configuration and/or state of the file systems may be re-created on FSVM-3. Host-2 200 b may continue to operate, in which case CVM-2 110 b may continue to execute on Host-2. When FSVM-2 again becomes available, e.g., after it has been re-started and has resumed operation, FSVM-2 may assert leadership and take back its IP address (10.1.1.2) and storage (FS1-Pool-3 366 b and FS2-Pool-4 367 b) from FSVM-1.
FIGS. 3G and 3H illustrate example virtualized file servers that have recovered from failure of a host machine 200 a by switching to another Controller/Service VM and another FSVM according to particular embodiments. The other Controller/Service VM and FSVM are located on a single host machine 200 c in FIG. 3G, and on two different host machines 200 b, 200 c in FIG. 3H. In both FIGS. 3G and 3H, Host-1 200 a has failed, e.g., crashed or otherwise become inoperative or unresponsive to network communication. Both FSVM-1 170 a and CVM-1 110 a located on the failed Host-1 200 a have thus failed. Note that the CVM 110 a and FSVM 170 a on a particular host machine 200 a may both fail even if the host machine 200 a itself does not fail. Recovery from failure of a CVM 110 a and an FSVM 170 a located on the same host machine 200 a, regardless of whether the host machine 200 a itself failed, may be performed as follows. The failure of FSVM-1 and CVM-1 may be detected by one or more other FSVMs, e.g., FSVM-2 170 b, FSVM-3 170 c, or by other components located on one or more other host machines. FSVM-1's failure may be detected when a communication timeout occurs or there is no response to a periodic status check message within a timeout period, for example. CVM-1's failure may be detected when a PATH DOWN condition occurs on one or more of CVM-1's volume groups' targets (e.g., iSCSI targets).
When FSVM-1's failure is detected, an election may be held as described above with reference to FIG. 3F to elect an active FSVM to take over leadership of the portions of the file server instance for which the failed FSVM was the leader. These portions are FileSystem-1A 364 a for the portion of file server FS1 located on FSVM-1, and FileSystem-2A 365 a for the portion of file server FS2 located on FSVM-1. FileSystem-1A 364 a uses the pool FS-Pool-1 366 a and FileSystem-2A 365 a uses the pool FS2-Pool-2 367 a. Thus, the FileSystem-1A 364 a and FileSystem-2A may be re-created on the new leader FSVM-3 170 c on Host-3 200 c. Further, FSVM-3 170 c may take over the IP address associated with failed FSVM-1 170 a, 10.1.1.1, so that storage access requests sent to FSVM-1 are received and processed by FSVM-3.
One or more failover paths from an FSVM to volume groups on one or more CVMs may be defined for use when a CVM fails. When CVM-1's failure is detected, the MPIO may activate one of the failover (e.g., standby) paths to remote iSCSI target volume group(s) associated with a remote CVM. For example, there may be a first predefined failover path from FSVM-1 to the volume groups VG1 368 c, 369 c in CVM-3 (which are on the same host as FSVM-1 when FSVM-1 is restored on Host-3 in examples of FIGS. 3G and 3H), and a second predefined failover path to the volume groups VG1 368 b, VG2 369 b in CVM-2. The first failover path, to CVM-3, is shown in FIG. 3G, and the second failover path, to CVM-2 is shown in FIG. 3H. An FSVM or MPIO may choose the first or second failover path according to the predetermined MPIO failover configuration that has been specified by a system administrator or user. The failover configuration may indicate that the path is selected (a) by reverting to the previous primary path, (b) in order of most preferred path, (c) in a round-robin order, (d) to the path with the least number of outstanding requests, (e) to the path with the least weight, or (f) to the path with the least number of pending requests. When failure of CVM-1 110 a is detected, e.g., by FSVM-1 or MPIO detecting a PATH DOWN condition on one of CVM-1's volume groups VG1 368 a or VG2 369 a, the alternate CVM on the selected failover path may take over I/O operations from the failed CVM-1. As shown in FIG. 3G, if the first failover path is chosen, CVM-3 110 c on Host-3 200 c is the alternate CVM, and the pools FS1-Pool-1 366 a and FS2-Pool-2 367 a, used by the file systems FileSystem-1A 364 a and FileSystem-2A 365 a, respectively, which have been restored on FSVM-3 on Host-3, may use volume groups VG1 368 c and VG2 369 c of CVM-3 110 c on Host-3 when the first failover path is chosen. Alternatively, as shown in FIG. 3H, if the second failover path is chosen, CVM-2 on Host-2 is the alternate CVM, and the pools FS1-Pool-1 366 a and FS2-Pool-2 367 a used by the respective file systems FileSystem-1A 364 a and FileSystem-2A 365 a, which have been restored on FSVM-3, may use volume groups VG1 368 b and VG2 369 b on Host-2, respectively.
Optionally, metadata used by FSVM-1 170 a, e.g., metadata associated with the file systems, may be transferred to FSVM-3 as part of the recovery process so that the specific configuration and/or state of the file systems may be re-created on FSVM-3. Further, metadata associated with the failed CVM-1 110 a, e.g., metadata related to volume groups 368 a, 369 a, may be transferred to the alternate CVM (e.g., CVM-2 or CVM-3) that the specific configuration and/or state of CVM-1 may be re-created on the alternative CVM. When FSVM-1 again becomes available, e.g., after it has been re-started and has resumed operation on Host-1 200 a or another host machine, FSVM-1 may assert leadership and take back its IP address (10.1.1.1) and storage assignments (FileSystem-1A and FS1-Pool-1 366 a, and FileSystem-2A and FS2-Pool-2 366 b) from FSVM-3. When CVM-1 again becomes available, MPIO or FSVM-1 may switch the FSVM to CVM communication paths (iSCSI paths) for FileSystem-1A 364 a and FileSystem-2A 365 a back to the pre-failure paths, e.g., the paths to volume groups VG1 368 a and 369 a in CVM-1 110 a, or the selected alternate path may remain in use. For example, the MPIO configuration may specify that fail back to FSVM-1 is to occur when the primary path is restored, since communication between FSVM-1 and CVM-1 is local and may be faster than communication between FSVM-1 and CVM-2 or CVM-3. In this case, the paths between CVM-2 and/or CVM-3 and FSVM-1 may again become failover (e.g., standby) paths.
FIGS. 4A and 4B illustrate an example hierarchical namespace 400 of a file server according to particular embodiments. Cluster-1 402 is a cluster, which may contain one or more file server instances, such as an instance named FS1.domain.com 404. Although one cluster is shown in FIGS. 4A and 4B, there may be multiple clusters, and each cluster may include one or more file server instances. The file server FS1.domain.com 404 contains three shares: Share-1 406, Share-2 408, and Share-3 410. Share-1 may be a home directory share on which user directories are stored, and Share-2 and Share-3 may be departmental shares for two different departments of a business organization, for example. Each share has an associated size in gigabytes, e.g., 100 GB (gigabytes) for Share-1, 100 GB for Share-2, and 10 GB for Share-3. The sizes may indicate a total capacity, including used and free space, or may indicate used space or free space. Share-1 includes three folders, Folder-A1 412, Folder-A2 414, and Folder-A3 416. The capacity of Folder-A1 is 18 GB, Folder-A2 is 16 GB, and Folder-A3 is 66 GB. Further, each folder is associated with a user, referred to as an owner. Folder-A1 is owned by User-1, Folder-A2 by User-2, and Folder-A3 by User-3. Folder-A1 contains a file named File-A1-1 418, of size 18 Gb. Folder-A2 contains 32 files, each of size 0.5 GB, named File-A2-1 420 through File-A2-32 422. Folder-A3 contains 33 files, each of size 2 GB, named File-A3-1 423 and File-A3-2 424 through File-A3-33 426.
FIG. 4B shows the contents of Share-2 408 and Share-3 410 of FS1.domain.com 404. Share-2 contains a folder named Folder-B1 440, owned by User-1 and having a size of 100 Gb. Folder-B1 contains File-B1-1 442 of size 20 Gb, File-B1-2 444 of size 30 Gb, and Folder-B2 446, owned by User-2 and having size 50 Gb. Folder-B2 contains File-B2-1 448 of size 5 Gb, File-B2-2 450 of size 5 Gb, and Folder-B3 452, owned by User-3 and having size 40 Gb. Folder-B3 452 contains 20 files of size 2 Gb each, named File-B3-1 454 through File-B3-20 456. Share-3 contains three folders: Folder-C7 429 owned by User-1 of size 3 GB, Folder-C8 430 owned by User-2 of size 3 GB, and Folder-C9 432 owned by User-3 of size 4 GB.
FIG. 4C illustrates distribution of stored data amongst host machines in a virtualized file server according to particular embodiments. In the example of FIG. 4C, the three shares are spread across three host machines 200 a-c. Approximately one-third of each share is located on each of the three FSVMs 170 a-c. For example, approximately one-third of Share-3's files are located on each of the three FSVMs 170 a-c. Note that from a user's point of a view, a share looks like a directory. Although the files in the shares (and in directories) are distributed across the three host machines 200 a-c, the VFS 202 provides a directory structure having a single namespace in which client executing on user VMs 101 and 102 may access the files in a location-transparent way, e.g., without knowing which host machines store which files (or which blocks of files).
In the example of FIG. 4C, Host-1 stores (e.g., is assigned to) 28 Gb of Share-1, including 18 Gb for File-A1-1 418 and 2 Gb each for File-A3-1 423 through File-A3-5 425, 33 Gb of Share-2, including 20 Gb for File-B1-1 and 13 Gb for File-B1-2, and 3 Gb of Share-3, including 3 Gb of Folder-C7. Host-2 stores 26 Gb of Share-1, including 0.5 Gb each of File-A2-1 420 through File-A2-32 422 (16 Gb total) and 2 Gb each of File-A3-6 426 through File-A3-10 427 (10 Gb total), 27 Gb of Share-2, including 17 Gb of File-B1-2, 5 Gb of File-B2-1, and 5 Gb of File-B2-2, and 3 Gb of Share-3, including 3 Gb of Folder-C8. Host-3 stores 46 GB of Share-1, including 2 GB each of File-A3-11 429 through File-A3-33 428 (66 GB total), 40 GB of Share-2, including 2 GB each of File-B3-1 454 through File-B3-20 456, and Share-3 stores 4 GB of Share-3, including 4 GB of Folder-C9 432.
In particular embodiments, a system for managing communication connections in a virtualization environment includes a plurality of host machines implementing a virtualization environment. Each of the host machines includes a hypervisor and at least one user virtual machine (user VM) 101. The system may also include a connection agent, an I/O controller, and/or a virtual disk comprising a plurality of storage devices. The virtual disk may be accessible by all of the I/O controllers, and the I/O controllers may conduct I/O transactions with the virtual disk based on I/O requests received from the user VMs 101. The I/O requests may be, for example, requests to perform particular storage access operations such as list folders and files in a specified folder, create a new file or folder, open an existing file for reading or writing, read data from or write data to a file, as well as file manipulation operations to rename, delete, copy, or get details, such as metadata, of files or folders. Each I/O request may reference, e.g., identify by name or numeric identifier, a file or folder on which the associated storage access operation is to be performed. The system further includes a virtualized file server, which includes a plurality of FSVMs 170 and associated local storage 122. Each FSVM 170 and associated local storage device 122 is local to a corresponding one of the host machines 200. The FSVMs 170 conduct I/O transactions with their associated local storage 122 based on I/O requests received from the user VMs 101. For each one of the host machines 200, each of the user VMs 101 on the one of the host machines 200 sends each of its respective I/O requests 383 to a selected one of the FSVMs 170, which may be selected based on a lookup table 360, e.g., a sharding map, that maps a file 318, folder 312, or other storage resource referenced by the I/O request to the selected one of the FSVMs 170).
In particular embodiments, the initial FSVM to receive the request from the user VM may be determined by selecting any of the FSVMs 170 on the network 140, e.g., at random, by round robin selection, or by a load-balancing algorithm, and sending an I/O request 383 to the selected FSVM 170 via the network 140 or via local communication within the host machine 200. Local communication may be used if the file 318 or folder 412 referenced by the I/O request is local to the selected FSVM, e.g., the referenced file or folder is located on the same host machine 200 as the selected FSVM 170. In this local case, the I/O request 383 need not be sent via the network 140. Instead, the I/O request 383 may be sent to the selected FSVM 170 using local communication, e.g., a local communication protocol such as UNIX domain sockets, a loopback communication interface, inter-process communication on the host machine 200, or the like. The selected FSVM 170 may perform the I/O transaction specified in the I/O request and return the result of the transaction via local communication. If the referenced file or folder is not local to the selected FSVM, then the selected FSVM may return a result indicating that the I/O request cannot be performed because the file or folder is not local to the FSVM. The user VM may then submit a REFERRAL request or the like to the selected FSVM, which may determine which FSVM the referenced file or folder is local to (e.g., by looking up the FSVM in a distributed mapping table), and return the identity of that FSVM to the user VM in a REDIRECT response or the like. Alternatively, the selected FSVM may determine which FSVM the referenced file or folder is local to, and return the identity of that FSVM to the user VM in the first response without the REFERRAL and REDIRECT messages. Other ways of redirecting the user VM to the FSVM of the referenced file are contemplated. For example, the FSVM that is on the same host as the requesting user VM (e.g., local to the requesting user VM) may determine which FSVM the file or folder is local to, and inform the requesting user VM of the identity of that FSVM without communicating with a different host.
In particular embodiments, the file or folder referenced by the I/O request includes a file server name that identifies a virtualized file server on which the file or folder is stored. The file server name may also include or be associated with a share name that identifies a share, file system, partition, or volume on which the file or folder is stored. Each of the user VMs on the host machine may send a host name lookup request, e.g., to a domain name service, that includes the file server name, and may receive one or more network addresses of one or more host machines on which the file or folder is stored.
In particular embodiments, as described above, the FSVM may send the I/O request to a selected one of the FSVMs. The selected one of the FSVMs may be identified by one of the host machine network addresses received above. In one aspect, the file or folder is stored in the local storage of one of the host machines, and the identity of the host machines may be determined as described below.
In particular embodiments, when the file or folder is not located on storage local to the selected FSVM, e.g., when the selected FSVM is not local to the identified host machine, the selected FSVM responds to the I/O request with an indication that the file or folder is not located on the identified host machine. Alternatively, the FSVM may look up the identity of the host machine on which the file or folder is located, and return the identity of the host machine in a response.
In particular embodiments, when the host machine receives a response indicating that the file or folder is not located in the local storage of the selected FSVM, the host machine may send a referral request (referencing the I/O request or the file or folder from the I/O request) to the selected FSVM. When the selected FSVM receives the referral request, the selected FSVM identifies one of the host machines that is associated with a file or folder referenced in the referral request based on an association that maps files to host machines, such as a sharding table (which may be stored by the centralized coordination service). When the selected FSVM is not local to the host machine, then the selected FSVM sends a redirect response that redirects the user VM on the host machine to the machine on which the selected FSVM is located. That is, the redirect response may reference the identified host machine (and by association the selected second one of the FSVMs). In particular embodiments, the user VM on the host machine receives the redirect response and may cache an association between the file or folder referenced in the I/O request and the host machine referenced in the redirect response.
In particular embodiments, the user VM on the host machine may send a host name lookup request that includes the name of the identified host machine to a name service, and may receive the network address of the identified host machine from the name service. The user VM on the host machine may then send the I/O request to the network address received from the name service. The FSVM on the host machine may receive the I/O request and performs the I/O transaction specified therein. That is, when the FSVM is local to the identified host machine, the FSVM performs the I/O transaction based on the I/O request. After performing or requesting the I/O transaction, the FSVM may send a response that includes a result of the I/O transaction back to the requesting host machine. I/O requests from the user VM may be generated by a client library that implements file I/O and is used by client program code (such as an application program).
Particular embodiments may provide dynamic referral type detection and customization of the file share path. Referring to FIG. 3C, when a user VM (e.g., client 330 or one of the user VMs 105 a-c) sends a request for a storage access operation specifying a file share to a FSVM node (e.g., FSVM-1 170 a) in the VFS cluster of FSVM nodes, the user VM may be sent a referral to another FSVM node (e.g., FSVM-3 170 c) that is assigned to the relevant file share. Certain types of authentication may use either host-based referrals (e.g., Kerberos) or IP-based referrals (e.g., NTLM). In order to flexibly adapt to any referral type, particular embodiments of the FSVM-1/3 170 a and 170 c may detect the referral type in an incoming request and construct a referral response that is based on the referral type and provide the referral. For example, if the user VM sends a request to access a storage item at a specified file share using an IP address, particular embodiments may construct and provide an IP address-based referral; if the user VM sends a request to access the storage item at the specified file share using a hostname, then particular embodiments may construct and provide a hostname-based referral, including adding the entire fully qualified domain name.
For example, if a user VM sends a request for File-A2-1 (which resides on Node-2, as illustrated in FIG. 4C) to Node-1 using a hostname-based address \\fs1\share-1\File-A2-1, VFS 202 may determine that File-A2-1 actually resides on Node-2 and send back a referral in the same referral type (hostname) as the initial request: \\fs2.domain.com\share-1\File-A2-1. If a user VM sends a request for File-A2-1 to Node-1 using an IP-based address \\198.82.0.23\share-1\File-A2-1, after determining that File-A2-1 actually resides on Node-2, VFS 202 may send back a referral in the same referral type (IP) as the initial request: \\198.82.0.43\share-1\File-A2-1.
In particular embodiments, the hostname for the referral node may be stored in a distributed cache in order to construct the referral dynamically using hostname, current domain, and share information.
FIG. 5 illustrates an example method for accessing data in a virtualized file server according to particular embodiments. The client system 330 may access the data, such as a specified folder, as follows. At step 502, the client system 330 receives a storage access request from an application executing in a user VM. Each storage access request references a file path (e.g., \\FS1.share.com\share-1\Folder-1), which includes a file or folder name and further includes or can be used to identify a share name (e.g., FS1.share.com\share-1) or an NFS remote file system name (e.g., fs1.share.com:/share-1. The storage access request may also include an operation type (e.g., read, write, delete, rename, etc.), a position in the file (for read/write requests), data to be written (for write requests), quantity of data to be read (for read requests), a new file path (for rename requests), folder name (for folder creation requests) or other information appropriate for the operation type. At step 504, the client system may send a DNS query request for the file server portion of the share name (e.g., \\fs1.domain.com for the share \\FS1.domain.com\share-1) to a name server 332, which may return the identity of a selected host machine as a result. The name server 332 may be a DNS server. The selected host machine is not necessarily the host machine on which the file or folder itself is located, however, since the share may be distributed amongst multiple host machines, one of which actually stores the file or folder. In particular embodiments, a FSVM each host machine can determine which host machine a file is stored on, and, if a FSVM receives a request for a file stored on a different host machine, the FSVM sends a referral response that includes the identity of the host machine on which the file is stored.
At step 506, the name server 332 may respond to the client with an IP (network) address of one or more host machines on FSVMs for the file or folder may be located. For example, the DNS server entry FS1.domain.com includes entries for FSVM-1, FSVM-2, and FSVM-3, which are, respectively, ip-addr1, ip-addr2, ip-addr3 (or 10.1.1.1, 10.1.1.2, 10.1.1.3). One of these three example IP addresses may be selected by the DNS server and returned in a response. In one example, the DNS server returns the three IP addresses in a different permutation for each request using DNS round robin so that a different server may be selected by the client for each request to balance the request load among the three servers. In this example, ip-addr1 (10.1.1.1) is the first address in the list sent in the reply to the client 330, and so is selected by the client as the address to which the I/O request will, at least initially, be sent. At step 508, the client may send the I/O request to access the folder “Folder-3” to the FSVM located on the host machine having address ip-addr1. The I/O request may be, e.g., a DFS attach or connect request, an NFS open request, or the like.
At step 510, FSVM-1 170 a on Host-1 200 a receives the I/O request and consults a map or lookup table, such as the sharding map 360 a, to determine whether the folder “Folder-3” is stored on a locally-attached storage resource of the host machine on which FSVM 170 a is located. If so, FSVM 170 a performs executes step 567 to perform the I/O transaction identified by the I/O request. If not, at step 512 FSVM-1 170 a responds to the client 330 with an indication that the folder is not located on the FSVM-1 170 a's host machine 200 a. The indication may be, e.g., a PATH_NOT_COVERED DFS response. At step 514, upon receiving the indication that the file is not located on the FSVM 170 a to which the request was sent, the client 330 sends a DFS REFERRAL request to FSVM 170 a, requesting a referral to the FSVM on which “Folder-3” is stored. At step 545, FSVM 170 a receives the REFERRAL request and sends a DFS “REDIRECT to FSVM-3” response back to the client 330. FSVM 170 a looks up the FSVM on which the folder “Folder-3” is stored in the map 360 a that associates files or shares with host machines. The result of the lookup, FSVM-3 170 c, may have been determined previously by the lookup at step 510 when the initial request for Folder-3 was received, or may be determined at step 516 when the referral request for Folder-3 is received. For example, the map 360 a may be stored in a shared data structure provided by the centralized coordination service, and the lookup may be performed by accessing the shared data structure. In this example, the file or folder is “Folder-3” and map indicates that the folder is associated with FSVM 170 c, so at step 516 FSVM 170 a may send a REDIRECT response to the client indicating that the requested folder is stored on host machine 200 c (on which FSVM 170 c is located). The REDIRECT response may reference the host machine 200 c, the FSVM 170 c, the network address of host machine 200 c (e.g., ip-addr3, in which case steps 518 and 520 may not be necessary), or other identifier for the location of the requested folder. The client 330 may receive the REDIRECT response and cache the association between Folder-3 and host machine 200 c (and/or FSVM 170 c) for potential future use.
At step 518, the client 330 may send a DNS query request to the DNS server 332 to determine the IP address of the FSVM specified in the received REDIRECT response, which is FSVM 170 c having IP address ip-addr3 in this example. At step 520, the DNS server 332 may send a reply to the client 330 indicating the IP address of the requested host machine. For example, the reply may be ip-addr3 (or 10.1.1.3), which is the IP address of FSVM 170 c. At step 522, the client sends the I/O request to access Folder-3 to the IP address received in the DNS reply (e.g., ip-addr3). At step 524, the FSVM 170 c on host machine 200 c receives the I/O request that references Folder-3 and looks up Folder-3 in the sharding map. At step 526, FSVM 170 c performs the requested I/O transaction for Folder-3, e.g., by accessing local storage 122 c, and sends the results of the access, e.g., details about Folder-3 in this example, such as a list of files and associated metadata, back to the client 330 in an I/O response. The client 330 receives the I/O response and may pass the results of the I/O transaction to the application or other program code that requested the access. Any subsequent requests for the same data (Folder-3 in this example) by the client 330 may be sent directly to host machine 200 c on which the data is stored because the client 330 may use the cached identity of the host machine or FSVM on which the data is stored. Although data contained in a folder is accessed in the example of FIG. 5 , other types of data may be accessed similarly, e.g., data contained in files.
In particular embodiments, a VFS 202 consists of multiple compute units, e.g., FSVMs 170. These FSVMs 170 act as a single VFS 202 to the outside world. Clusters with appropriate platforms and licenses use a hypervisor-agnostic code-image for the VFS 202. This image may be stored as part of pre-created, ready to use disk images. When a user with correct privileges decides to create a VFS 202, the image may be cloned N times, where N is the number of FSVMs 170, and the FSVMs 170 are created. The FSVMs 170 form a cluster, which may provide a VFS 202 to the outside world. In this way, the user is abstracted from the complex process of deploying the VFS 202, as the input requested from the user is a small number of parameters that can be provided by the user in a simple user interface. The pre-created fileserver image may reduce the deployment time to be as fast as booting the host machines 200.
In particular embodiments, the VFS comprises multiple FSVMs 170. The host machines 200 may combine to act as VFS 202 to the outside world. Each host machine 200 may have two types of vDisks: code and data. The operating system (OS) code and file server code reside on the code vDisk. The fileserver persistent data and configuration are stored on the data vDisk. In a first technique for upgrading the VFS 202, before the upgrade process is started, the newer version of the code disk is prepared and cloned N times (where N is the number of FSVMs 170). While upgrading the VFS 202 to the latest version, a new code disk is swapped with the existing code disk for each FSVM 170. After rebooting the FSVM 170, it will be running with newer code, and continues serving the data using the newer code.
In particular embodiments, in a second technique for upgrading the VFS 202, before the upgrade process is started and after the newer version of the code disk is prepared and cloned, a first FSVM 170 a acquires an upgrade token, swaps the old code disk with the newer disk, and reboots. When the first FSVM 170 a comes back up and is running, the upgrade token is passed to the next FSVM 170 b, which may perform the swap and reboot, and pass the upgrade token to the next FSVM 170 c. These operations are repeated until the last FSVM 170, e.g., FSVM 170 c in this example, is upgraded. During the time that each FSVM 170 b is being rebooted, one of the peer FSVMs 170 a takes over the storage and IP address of the FSVM 170 b so that the client does not see any interruption in the file service.
In particular embodiments, users dealing with Virtual Disk Image (“VDI”) files and their corresponding root directories are, by definition, bound to their VMs. This binding provides a user VM 101 to root-level directory mapping. A sharding algorithm may determine a mapping between a user VM 101 and its corresponding host machine 200. This mapping may in turn provide a root-level directory-to-host machine mapping. The sharding algorithm may use this mapping and add metadata to keep storage units and compute units local, e.g., located on, the same host machine. On migration of the user virtual machines 102, metadata and storage will be moved accordingly.
Particular embodiments may provide enhanced performance via adaptive data and compute-unit migration. Particular embodiments may provide the ability to restrict compute units and storage units to a location governed by user policy.
In particular embodiments, data migration from an existing VFS 202 to a new VFS 202 may be bounded by the speed of connection between the existing infrastructure (e.g., host machines 200) and the new system (e.g., other host machines). By using smart data ingestion, data migration speed can be increased with a multiplier of the number of file server host machines.
In previous approaches, data is migrated using a utility to copy data from one source to one target location. Migration speed is limited by the connection speed. In particular embodiments, using the smart data ingestion approach described herein, top-level directories in an existing VFS 202 are preprocessed to acquire the destination host machine 200 on a new (destination) VFS 202. When data migration begins, each host machine 200 in the VFS 202 starts data migration with the share directories assigned, which speeds up data migration with a multiplier of host machine count. By taking advantage of the distributed nature of the VFS 202, data migration is performed in parallel to speed up the migration process. Using the same sharding algorithm as file I/O to decide the migration target ensure the consistency of migrated data placement in, e.g., the new (destination) VFS 202. In particular embodiments, no further processing is needed after data is migrated, and data is ready to be served.
In a first example, when the organization that manages a virtualized file server instance (VFS) 202 decides to, for example, segregate the existing VFS 202 to a departmental level, the VFS 202 may be split into multiple virtualized file server instances (VFSs) 202 without affecting the stored data, with zero to minimal down time, and with zero data copying or migration. In a second example, when an organization that manages multiple VFSs 202 decides to merge them into one manageable VFS 202, the multiple VFSs 202 may be merged together without affecting the stored data, and with zero to minimal down time and zero data copying or migration. When an organization needs to merge multiple VFSs 202, then a system administrator may deploy a new VFS 202 and migrate the stored data from the multiple VFSs 202 to the newly deployed VFS 202, which takes more time and resources. When the organization needs to split the VFS 202, then a system administrator may deploy new VFSs 202 and migrate the data from the old VFS 202 to the newly deployed VFSs 202, which also takes more time and resources.
In particular embodiments, the splitting and merging operations may be performed as follows. To split an existing VFS 202, e.g., upon a system administrator's request, the following operations may be performed:
1. Select the FSVMs 170 to be segregated from VFS 202.
2. The FSVMs 170 are removed one by one.
3. Before removing a FSVM 170 from the VFS 202, first select a lightly loaded FSVM 170 and voluntarily relinquish the storage resources to the selected FSVM 170. The IP address of the FSVM 170 being removed may also be moved to the selected FSVM 170 to retain SMB client connections.
4. After removing all the FSVMs 170 from the VFS 202, a new VFS 202 is constructed.
5. The FSVMs 170 of the new VFS 202 join the domain and start serving the new shares. Old shares may still be served by the old VFS 202. Once the administrator decides to move the old shares to the new VFS 202, trigger a storage transition that relinquishes the storage to the appropriate selected FSVMs 170 and move the IP addresses of FSVMs 170 of the old VFS 202 to FSVMs 170 of the new VFS 202.
6. The same process may be continued to segregate other VFSs 202.
In particular embodiments, to merge multiple VFSs 202 together, e.g., upon a system administrator's request, an election is triggered between the multiple VFSs 202 based on the virtual IP address or based on preference policies. The VFS 202 that wins the election or is selected by an administrator is treated as a master VFS. All other VFSs then join to the master VFS's ACTIVE DIRECTORY domain. FSVMs 170 from all slave VFSs 202 may be added to the master VFS 202, and storage pool metadata of the slave VFSs 202 is modified to serve for the master VFS 202. The following operations may be performed to merge the slave VFSs into the master VFS:

- 1. Select the VFSs 202 to be merged.
- 2. Initiate the election to elect the master VFS 202 based on the policy of the VFS 202 that has the higher IP address.
- 3. Once the master VFS has been selected, clients connect to it.
- 4. Select a slave VFS to merge.
- 5. Relinquish the storage to a lightly-loaded FSVM on the master VFS and move the IP address to refer to the lightly-loaded FSVM.
- 6. Start serving SMB clients for new and old shares.
- 7. Stop the slave file server, add its FSVM(s) one by one to the new master file server, and take back its resource on the new master file server.
- 8. Continue these steps for other slave file servers.

In scenarios such as a company splitting into multiple companies, it could be a requirement that a single VFS 202 is split into two VFSs 202. However, there may be certain SAMBA shares in the original VFS 202 that need to be made accessible to both the VFSs 202. As an example, consider two different fileservers FS1 and FS2. FS1 originally hosted a share ‘share1’ FS2 needs the ability to read/write to the share ‘share1’. The SMB requests for ‘share1’ on FS2 may be forwarded or proxied to FS1, thereby allowing the share ‘share1’ to be readable/writable from two different VFSs 202. Another approach is to NFS-mount the original share on the new VFS 202, to provide a single namespace. The ability to selectively choose certain shares to be shared across other VFSs 202 ensures a tight security boundary at the VFS level, along with the collaborative access via two different VFSs.
In particular embodiments, disaster recovery of the VFS 202 may be performed by making backups and replicating delta changes in a storage layer, then recovering the data stored in the VFS 202 at a remote site. The data may be recovered by reconstructing the VFS 202 from a replicated configuration. In a production environment, the data stored on a VFS 202 is securely protected and restored on a remote location without loss of the data and metadata within a supported Recovery Point Objective (which may be the age of files to be recovered from backup storage for normal operations to resume). A custom replication policy may be configured for the VFS 202, and the ability may be provided to map the VFS 202's configuration between sites to provide disaster recovery of virtual file-services across geographical locations. Particular embodiments may provide the ability to protect individual shares or share groups by protecting the volume group(s) used for file-services storage, e.g., by adding them to a protection domain. Users may apply the replication and backup policies on the protection domain to configure the Recovery Point Objective, recovery sites (alternate cluster or cloud), and replication constraints such as bandwidth and schedule. Particular embodiments may take lightweight snapshots and transfer the delta of the snapshots for the given volume groups. Along with file-services share data, particular embodiments may also transfer the VFS configuration e.g. file-server size, compute-unit configuration, and metadata, e.g., share ACLs, quotas, and so on. Particular embodiments may also provide a simplified user interface to configure mapping of network, DNS-servers, active-directory, etc. between remote sites. Potential benefits may include:

- 1. Granular level of protection (share or group of shares) to configure different Recovery Point Objective.
- 2. Custom replication policies to utilize the network resources effectively for replication.
- 3. Fine control on network and ecosystem resource mapping between sites.
- 4. Light weight snapshot includes share data delta, metadata and file-server configuration leading to less replication traffic across sites.
- 5. One click restore of file-services on remote site.
- 6. Distribution of share replication across multiple remote sites.
- 7. Multiple recovery points on multiple remote sites for multi-site failures.

In particular embodiments, a high volume of snapshots (particularly lightweight snapshots) may be stored, and in some cases, snapshots may be captured and stored using a timestamp format that differs from the naming convention used by the client to access the snapshot. This may cause significant latency when browsing through thousands of snapshots, due to the performance overhead required to read the stat or query the file system to get the snapshot create time and then map it to corresponding client side label. For example, the underlying file system may choose to store snapshots using afs-auto-snapshot-hourly-YYYY-MM-DD-HHMM naming format; if the user VMs are then using SMB requests to retrieve the snapshots, they may be using @GMT tokens to access the snapshots. @GMT tokens are special token that can be present as part of a file path to indicate a request to see a previous version of the file or directory. The format is “@GMT-YYYY.MM.DD-HH.MM.SS”.
Particular embodiments may, at the time of capturing and/or storing a snapshot, extract the creation timestamp from the snapshot name, utilize the creation timestamp to determine the expected request format then cache a mapping from the expected request format to the actual name of the corresponding snapshot. Latency may thereby be significantly reduced to a one-time computation that takes place at capture-time, rather than potentially multiple system calls taking place at recovery time.
FIG. 6 illustrates an example of how a file server ‘FS1’ may be deployed across multiple clusters according to particular embodiments. Particular embodiments may facilitate deploying and managing a VFS 202 whose networking, compute-unit, and storage resources are distributed across multiple clusters from a single management portal. Particular embodiments may create a VFS 202 and distribute compute units, which may be the FSVMs 170. A portal user interface may be used by a user or administrator to create a VFS 202. While creating the VFS 202, a user is given a list of clusters that may be used to distribute the compute units (e.g., FSVMs, or may perform the operations of FSVMs as described herein), networking (IP addresses) and storage (containers). In the example of FIG. 6 , the user has chosen three clusters, Cluster 1, Cluster 2, and Cluster 3. Three FSVMs are created on each cluster, for a total of 9 FSVMs across the three clusters. Each cluster for this file server hosts a separate container, which holds a part of the file server data. The containers are labeled Container 1, Container 2, and Container 3. The containers are hidden from the user.
Particular embodiments may create shares and distribute storage units and compute units. The portal user interface may be used to create a share ‘share1’ within the file server FS1. The data within ‘share1’ is distributed across all the clusters. A storage pool of multiple vDisks is constructed on all the FSVMs across all the clusters. Each storage pool on each FSVM is responsible for a subset of the ‘share1’ data. The share is sharded at the top-level directories across FSVMs residing in different clusters. The sharding strategy is as follows. Assuming that directories dir1, dir2, dir3, dir4, dir5, dir6 have been created:

- 1. Each FSVM within each cluster hosts a storage pool created from a subset of the container storage. A background process periodically runs on a leader FSVM in each cluster to aggregate the File system space used for each share across all FSVMs in the cluster. This data is published to a cluster manager that stores the data in an entity database, e.g., APACHE CASSANDRA or the like. The cluster manager may be NUTANIX PRISM CENTRAL, which is a multi-cluster manager responsible for managing multiple clusters to provide a single, centralized management interface.
- 2. User creates a new top-level directory, e.g., ‘dir7’.
- 3. The Samba VFS layer intercepts the directory creation request and consults a database to determine whether the directory is hosted by any FSVM (or FSVM). If it is not, the VFS layer makes an RPC call to a file server service running in PRISM CENTRAL to identify a location (which may be an optimal location) for dir7′.
- 4. The file server service running in PRISM CENTRAL retrieves the per-cluster usage statistics for each share that it received in step 1 above, and chooses the cluster that has the least used space for the share ‘share1’. In the example of FIG. 6 , Cluster 1 is chosen. The file server service may also provide an option to simply choose the cluster that has the greatest amount of free fileserver container space.
- 5. Next, the file server service running in PRISM CENTRAL queries Cluster 1 for average CPU utilization for the past 24 hours for all VMs within Cluster 1. The file server service then chooses the least loaded FSVM. The file server service in PRISM CENTRAL returns this <Cluster 1, FSVM2 FQDN> tuple back to the VFS layer.
- 6. The VFS layer now knows the FSVM2 FQDN, which should host dir7′ and hence creates this new directory on the Unix file system corresponding to FSVM2. The VFS layer records this mapping <Share1, dir1>→Cluster 1, FSVM2 in a database, and returns a PATH_NOT_COVERED message to the client.
- 7. Through DFS referral, the SAMBA client requests the path for the directory ‘dir1’. The FSVM looks up ‘dir1’ in the database, and returns FSVM2 IP to the client. The client now accesses ‘dir1’ on the FSVM2 file system.

The file system on any FSVM may be composed of vDisks. Since vDisks are distributed across the cluster, this arrangement provides uniform sharding of storage within the cluster. The sharding strategy described above causes all clusters' containers and FSVMs to be used, and achieves uniform sharding of storage units and compute units across the clusters.
Particular embodiments may provide cluster-aware sharding and cluster-aware share level quotas. At the time of share creation, user is given the option to co-locate the data for the share within certain clusters. This option may be useful if the user wishes to have one set of shares distributed within a certain geographical boundary, and a different set of shares distributed across a different geographical boundary, in which case the above sharding strategy remains the same. In step 4 above, only those clusters that were selected while creating the share would be made available to be considered for sharding. This technique provides cluster-aware sharding.
Similarly, quotas can be set on a file server service 202. Quotas may set a limit on the amount of data storage to be used for each share within each cluster. Since file server service stores a per-share, per-cluster storage usage, it can detect when a cluster-level share quota is reached. Depending on the quota policy, the user may be alerted when this cluster-level quota is reached, or the file server service may notify the FSVM (or FSVM) leader within the cluster whose quota has been reached via RPC. On receiving this notification, the FSVM leader may make all file systems for that share across FSVMs read-only to respect the storage quota limit.
Particular embodiments may handle sharding by considering geographic quotas, user-based quotas, fault tolerance of clusters, available resources across clusters, etc. Some benefits may include:

- 1. Provides uniform auto sharding of compute, network, and storage units across multiple clusters, which also leads to a smaller fault domain
- 2. File server centrally managed from a single interface, although resources are distributed, leading to easy manageability.
- 3. Provides flexibility of co-locating shares where necessary and distributing them across clusters when necessary.
- 4. Provides ability to set cluster aware share level quotas, which could be utilized for location-aware sharding. and
- 5. Fault tolerant within cluster and capable of tolerating entire cluster failure.

Particular embodiments may identify corrupted or infected data and recover a consistent version of the data from a VFS 202. When user data is infected by a virus or corrupted by a file system or storage system, identifying the corrupted data and the needed recovery level may be difficult. If the appropriate recovery level is not detected and data is recovered at the wrong place, then a valid version of data may be lost. Particular embodiments may provide capabilities to virtual file services to detect problems from file level to storage level. System administrators need not worry about detecting and recovering a consistent version of data when the system administrator detects the corruption and infected data and manually recovers the data from a file system or from storage system. A self-healing mechanism of the VFS 202 frequently takes snapshots of file system and storage pools and monitors the user data at file system and storage system levels. In particular embodiments, a virtualized file server may accomplish the following levels of detection and recovery:

- 1. File/folder level recovery: File system or anti-virus or other internal modules can detect the file or folder level infection or corruption. Self-Healing mechanism monitors these events and once it detects, it will recover those particular data from the previous file system snapshot by overwriting the infected/corrupted files/folders.
- 2. File System level recovery: Self-healing mechanism monitors the checksum of the file system and if it finds any discrepancy on that, it will recover the file system with its latest snapshot.
- 3. Storage level recovery: Self-Healing mechanism monitors storage-pool corruption and alerts generated by the cluster and detect the data loses and corruption. Once it detects the data corruption/data loss, it will recover the storage-pool for the latest snapshot.

Distributed Self-Healing: Since virtualized file server compute and storage units are distributed across multiple host machines 200, the self-healing mechanism efficiently monitors the corruptions and data loss in parallel and distributed fashion on all the host machines and detects and recovers that particular data without affecting the overall file server 202.
Some benefits may include: Detection of and recovery from data loss, data corruption and infected files on file/folder level, file system level and storage level without manual intervention. Efficient detection of and recovery from the data loss, data corruption and infected files in parallel and distributed fashion. Recovery from data loss and data corruption without affecting the overall file server 202.
Particular embodiments may back up cold data stored in a cluster to an object store, which is either in a public cloud (e.g., AMAZON WEB SERVICES), or to a low-cost storage medium within the same cluster. Particular embodiments may then retrieve the backed-up volume groups as needed to restore files for the file server. Particular embodiments may provide a method to backup data on a virtualized file server running on a hyper-converged infrastructure to a low-cost storage medium hosted on the same physical infrastructure. This consists of a virtualized server running on the same hyper-converged infrastructure providing an object store interface (such as AMAZON S3) with storage as low-cost media such as SMR drives. This particular virtual machine can act as a backup server for other VMs running on the same infrastructure.
Particular embodiments of the backup server may be hosted on the same hyper converged infrastructure as the compute and storage. Particular embodiments of the backup server may be used for low cost storage media like SMR drives attached to the same hyper converged infrastructure. Particular embodiments of the backup server may provide generic object store interfaces such as AMAZON S3. Particular embodiments of the backup server may provide the same level of availability as the other highly available services (such as FSVM) run on the cluster.
Particular embodiments may include a cloud service as a storage tier of a virtualized file server service. Particular embodiments may then retrieve the backed-up volume groups as needed to restore files for the file server.
Particular embodiments may provide block awareness for a virtualized file server service in order to maintain availability of virtual file server services in case of block failure by deploying FSVMs 170 on different host machines 200. In case of block failure (e.g., due to power loss affecting a block, a.k.a. hardware host machine), the high-availability features attempt to migrate the VMs running on those host machines to available running host machines. If there are not enough resources on the available running host machines, then the file-server HA features are triggered, and online FSVM(s) take ownership of the volume-group of offline FSVM(s). When one FSVM with metadata service is down, the file-server may continue to serve requests to end users without any performance degradation. Potential benefits may include:

- 1. Virtualized file-server is available even if one block in the cluster goes down.
- 2. Users or administrators need not to reserve the resources or free up the resources for block failure to get the virtualized file-server working.
- 3. In a hyper-converged deployment, the user VMs can be prioritized over FSVMs for migration during block failure.

Particular embodiments may recover from multimode file service failures in a scale-out NAS environment with minimal down time. Traditional file server deployments protected against single host machine failures by having a standby host machine. Detection of service failures is not spontaneous and issues can occur with keeping backup host machines synchronized. Further, if the new active host machine is also down, there may be no way to recover the service. These issues not only cause service interruption but also create complications if there are multiple host machine failures.
In a scale out NAS environment, host machine and service failures may be detected and recovered from without interrupting service to clients. Using distributed cluster health service, these service failures may be detected, and the other active host machines may take over the failed host machine services (both storage and network). Each host machine in cluster acts as potential backup host machine, which will help with managing multiple simultaneous host machine failures based on cluster health. So, even if the new active host machine is down, other host machines in the cluster can take over the new active host machine's load and provide continuous availability. In this way, clients using the scale-out NAS services do not see any downtime or service interruptions when multiple host machines in the cluster are down.
Particular embodiments may help to avoid catastrophic failures due to resource exhaustion. In scenarios such as a user's home directory being accessed for read/write operations, the user may not be able to determine how much disk quota is assigned to the user or how much actual space is available to the user to write data.
As an example, consider a scenario when many users have their home directories on the same share. Existing technologies display the user's drive size as being the same as total share size, thereby giving the user the perception that the total share size is available to write data. However, when the user's quota limit has been met or exceeded, any write to the drive fails.
Particular embodiments may address the administrative challenge of creating and managing a large number of home directories for users. Typically, a home directory is created and maintained for each user account. Conventional file systems place all the home directories on a dedicated home share comprising storage residing on a specific node (e.g., a single host machine storing a volume group \\home containing home directories \\home\user-A, \\home\user-B, . . . , \\home\user-N)—this may become a challenge to manage when there are many users, requiring a large number of shares. Certain file systems may distribute the home directories across multiple nodes in the cluster and create separate home shares (e.g., a first host machine storing a volume group \\home1 containing home directories \\home1\user-A, \\home1\user-B, . . . , \\home1\user-F, and a second host machine storing a volume group \\home2 containing home directories \\home2\user-G, \\home2\user-H, . . . , \\home2\user-N).
In particular embodiments, certain directories may be designated as home shares, and user home directories may be created as top-level directories in a distributed home share. The home directories may be distributed across a plurality of nodes while providing the ability to serve storage access operations addressed to the single distributed home share (e.g., a request to access a file in a user directory that is sent to a first FSVM running on a first host machine may be sent a share-level DFS referral to a second host machine, where the user directory resides). Particular embodiments may create home directories for users on multiple different host machines. Upon successfully authenticating a user, the user's home directory is exported. When the client sends a request to perform a storage access operation to a first FSVM, if the user's home directory resides in local storage of the host machine on which the first FSVM is running, then the FSVM sends a response to the storage access operation. If the user's home directory resides in remote storage, then the FSVM may force a lookup to determine which host machine is storing the user's home directory in its local storage, construct a reference to the actual location of the user's home directory, then send back a share-level DFS referral to redirect future requests to perform storage access operations in the user's home directory to the correct host machine.
In particular embodiments, if a home directory does not yet exist for a new user, the home directory for the user may be created in local storage for the host machine running the FSVM to which the request was sent. If the home directory for the user needs to be created on a different host machine (e.g., due to load balancing or capacity issues), particular embodiments, may create the new home directory on another host machine, then send back a share-level DFS referral to that host machine.
In particular embodiments, home directories may be redistributed on the fly. For example, in a situation where each of three host machines A, B, and C stores home directories for 20, 20, and 15 users respectively, home machine A may be experiencing undue strain on its resources (e.g., with respect to consumption of processor cycles, due to a high frequency of I/O transactions, or with respect to disk usage, due to a large number of files that are being stored). Once a determination is made that the home directories on host machine A that are triggering high resource consumption need to be redistributed, the data from such home directories may be copied over to a new host machine (e.g., host machine C), and once the data is completely or mostly copied over, FSVMs may start redirecting requests relating to the moved home directories to host machine C by way of sending back share-level DFS referrals.
Particular embodiments may expose user-specific data so that when each user accesses their VDI environment they see their home directory as a mounted drive, and see data specific to their disk portion, such as disk capacity, average rate of utilization of space, frequency of disk accesses, file type, etc. On every soft quota limit reached, the user may be alerted through email that they are about to exhaust their disk quota. Less-frequently-accessed files, folders, and other items may be archived automatically to the cloud.
Particular embodiments may provide high availability of storage services in a scale out file-server. In traditional file server deployments, high-availability is supported by configuring host machines as pairs where the storage resources are inter-connected between two host machines. So, if one of the host machines fails, the other host machine in the pair may take over the storage resources along with the IP address. One limitation with this approach is that an even number of host machines is needed in the cluster. In a scale out file-server, minimal to zero disruption occurs in case of any failure. In a virtualized scale-out file server, all host machines in the scale out cluster monitor the health for every other host machine. If one of the host machines experiences down time because of either a planned shutdown or unplanned host machine failures, one of the host machines starts taking over the storage resources of the down host machine. At the same time, the IP address fails over so that clients can continue to contact the takeover host machine without any disruptions. To increase the load balancing, the failover storage resources may be distributed to multiple host machines, so that the down host machine resources may be distributed across different host machines.
FIG. 7 is a block diagram of an illustrative computing system 700 suitable for implementing particular embodiments. In particular embodiments, one or more computer systems 700 perform one or more steps of one or more methods described or illustrated herein. In particular embodiments, one or more computer systems 700 provide functionality described or illustrated herein. In particular embodiments, software running on one or more computer systems 700 performs one or more steps of one or more methods described or illustrated herein or provides functionality described or illustrated herein. Particular embodiments include one or more portions of one or more computer systems 700. Herein, reference to a computer system may encompass a computing device, and vice versa, where appropriate. Moreover, reference to a computer system may encompass one or more computer systems, where appropriate.
This disclosure contemplates any suitable number of computer systems 700. This disclosure contemplates computer system 700 taking any suitable physical form. As example and not by way of limitation, computer system 700 may be an embedded computer system, a system-on-chip (SOC), a single-board computer system (SBC) (such as, for example, a computer-on-module (COM) or system-on-module (SOM)), a desktop computer system, a mainframe, a mesh of computer systems, a server, a laptop or notebook computer system, a tablet computer system, or a combination of two or more of these. Where appropriate, computer system 700 may include one or more computer systems 700; be unitary or distributed; span multiple locations; span multiple machines; span multiple data centers; or reside in a cloud, which may include one or more cloud components in one or more networks. Where appropriate, one or more computer systems 700 may perform without substantial spatial or temporal limitation one or more steps of one or more methods described or illustrated herein. As an example and not by way of limitation, one or more computer systems 700 may perform in real time or in batch mode one or more steps of one or more methods described or illustrated herein. One or more computer systems 700 may perform at different times or at different locations one or more steps of one or more methods described or illustrated herein, where appropriate.
Computer system 700 includes a bus 702 (e.g., an address bus and a data bus) or other communication mechanism for communicating information, which interconnects subsystems and devices, such as processor 704, memory 706 (e.g., RAM), static storage 708 (e.g., ROM), dynamic storage 710 (e.g., magnetic or optical), communication interface 714 (e.g., modem, Ethernet card, a network interface controller (NIC) or network adapter for communicating with an Ethernet or other wire-based network, a wireless NIC (WNIC) or wireless adapter for communicating with a wireless network, such as a WI-FI network), input/output (I/O) interface 712 (e.g., keyboard, keypad, mouse, microphone). In particular embodiments, computer system 700 may include one or more of any such components.
In particular embodiments, processor 704 includes hardware for executing instructions, such as those making up a computer program. As an example and not by way of limitation, to execute instructions, processor 704 may retrieve (or fetch) the instructions from an internal register, an internal cache, memory 706, static storage 708, or dynamic storage 710; decode and execute them; and then write one or more results to an internal register, an internal cache, memory 706, static storage 708, or dynamic storage 710. In particular embodiments, processor 704 may include one or more internal caches for data, instructions, or addresses. This disclosure contemplates processor 704 including any suitable number of any suitable internal caches, where appropriate. As an example and not by way of limitation, processor 704 may include one or more instruction caches, one or more data caches, and one or more translation lookaside buffers (TLBs). Instructions in the instruction caches may be copies of instructions in memory 706, static storage 708, or dynamic storage 710, and the instruction caches may speed up retrieval of those instructions by processor 704. Data in the data caches may be copies of data in memory 706, static storage 708, or dynamic storage 710 for instructions executing at processor 704 to operate on; the results of previous instructions executed at processor 704 for access by subsequent instructions executing at processor 704 or for writing to memory 706, static storage 708, or dynamic storage 710; or other suitable data. The data caches may speed up read or write operations by processor 704. The TLBs may speed up virtual-address translation for processor 704. In particular embodiments, processor 704 may include one or more internal registers for data, instructions, or addresses. This disclosure contemplates processor 704 including any suitable number of any suitable internal registers, where appropriate. Where appropriate, processor 704 may include one or more arithmetic logic units (ALUs); be a multi-core processor; or include one or more processors 702. Although this disclosure describes and illustrates a particular processor, this disclosure contemplates any suitable processor.
In particular embodiments, I/O interface 712 includes hardware, software, or both, providing one or more interfaces for communication between computer system 700 and one or more I/O devices. Computer system 700 may include one or more of these I/O devices, where appropriate. One or more of these I/O devices may enable communication between a person and computer system 700. As an example and not by way of limitation, an I/O device may include a keyboard, keypad, microphone, monitor, mouse, printer, scanner, speaker, still camera, stylus, tablet, touch screen, trackball, video camera, another suitable I/O device or a combination of two or more of these. An I/O device may include one or more sensors. This disclosure contemplates any suitable I/O devices and any suitable I/O interfaces 712 for them. Where appropriate, I/O interface 712 may include one or more device or software drivers enabling processor 704 to drive one or more of these I/O devices. I/O interface 712 may include one or more I/O interfaces 712, where appropriate. Although this disclosure describes and illustrates a particular I/O interface, this disclosure contemplates any suitable I/O interface.
In particular embodiments, communication interface 714 includes hardware, software, or both providing one or more interfaces for communication (such as, for example, packet-based communication) between computer system 700 and one or more other computer systems 700 or one or more networks. As an example and not by way of limitation, communication interface 714 may include a network interface controller (NIC) or network adapter for communicating with an Ethernet or other wire-based network or a wireless NIC (WNIC) or wireless adapter for communicating with a wireless network, such as a WI-FI network. This disclosure contemplates any suitable network and any suitable communication interface 714 for it. As an example and not by way of limitation, computer system 700 may communicate with an ad hoc network, a personal area network (PAN), a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), or one or more portions of the Internet or a combination of two or more of these. One or more portions of one or more of these networks may be wired or wireless. As an example, computer system 700 may communicate with a wireless PAN (WPAN) (such as, for example, a BLUETOOTH WPAN), a WI-FI network, a WI-MAX network, a cellular telephone network (such as, for example, a Global System for Mobile Communications (GSM) network), or other suitable wireless network or a combination of two or more of these. Computer system 700 may include any suitable communication interface 714 for any of these networks, where appropriate. Communication interface 714 may include one or more communication interfaces 714, where appropriate. Although this disclosure describes and illustrates a particular communication interface, this disclosure contemplates any suitable communication interface.
One or more memory buses (which may each include an address bus and a data bus) may couple processor 704 to memory 706. Bus 702 may include one or more memory buses, as described below. In particular embodiments, one or more memory management units (MMUs) reside between processor 704 and memory 706 and facilitate accesses to memory 706 requested by processor 704. In particular embodiments, memory 706 includes random access memory (RAM). This RAM may be volatile memory, where appropriate. Where appropriate, this RAM may be dynamic RAM (DRAM) or static RAM (SRAM). Moreover, where appropriate, this RAM may be single-ported or multi-ported RAM. This disclosure contemplates any suitable RAM. Memory 706 may include one or more memories 706, where appropriate. Although this disclosure describes and illustrates particular memory, this disclosure contemplates any suitable memory.
Where appropriate, the ROM may be mask-programmed ROM, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), electrically alterable ROM (EAROM), or flash memory or a combination of two or more of these. In particular embodiments, dynamic storage 710 may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc, a magneto-optical disc, magnetic tape, or a Universal Serial Bus (USB) drive or a combination of two or more of these. Dynamic storage 710 may include removable or non-removable (or fixed) media, where appropriate. Dynamic storage 710 may be internal or external to computer system 700, where appropriate. This disclosure contemplates mass dynamic storage 710 taking any suitable physical form. Dynamic storage 710 may include one or more storage control units facilitating communication between processor 704 and dynamic storage 710, where appropriate.
In particular embodiments, bus 702 includes hardware, software, or both coupling components of computer system 700 to each other. As an example and not by way of limitation, bus 702 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a front-side bus (FSB), a HYPERTRANSPORT (HT) interconnect, an Industry Standard Architecture (ISA) bus, an INFINIBAND interconnect, a low-pin-count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCIe) bus, a serial advanced technology attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or another suitable bus or a combination of two or more of these. Bus 702 may include one or more buses 706, where appropriate. Although this disclosure describes and illustrates a particular bus, this disclosure contemplates any suitable bus or interconnect.
According to particular embodiments, computer system 700 performs specific operations by processor 704 executing one or more sequences of one or more instructions contained in memory 706. Such instructions may be read into memory 706 from another computer readable/usable medium, such as static storage 708 or dynamic storage 710. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions. Thus, particular embodiments are not limited to any specific combination of hardware circuitry and/or software. In one embodiment, the term “logic” shall mean any combination of software or hardware that is used to implement all or part of particular embodiments disclosed herein.
The term “computer readable medium” or “computer usable medium” as used herein refers to any medium that participates in providing instructions to processor 704 for execution. Such a medium may take many forms, including but not limited to, nonvolatile media and volatile media. Non-volatile media includes, for example, optical or magnetic disks, such as static storage 708 or dynamic storage 710. Volatile media includes dynamic memory, such as memory 706.
Common forms of computer readable media include, for example, floppy disk, flexible disk, hard disk, magnetic tape, any other magnetic medium, CD-ROM, any other optical medium, punch cards, paper tape, any other physical medium with patterns of holes, RAM, PROM, EPROM, FLASH-EPROM, any other memory chip or cartridge, or any other medium from which a computer can read.
In particular embodiments, execution of the sequences of instructions is performed by a single computer system 700. According to other particular embodiments, two or more computer systems 700 coupled by communication link 716 (e.g., LAN, PTSN, or wireless network) may perform the sequence of instructions in coordination with one another.
Computer system 700 may transmit and receive messages, data, and instructions, including program, i.e., application code, through communication link 716 and communication interface 714. Received program code may be executed by processor 704 as it is received, and/or stored in static storage 708 or dynamic storage 710, or other non-volatile storage for later execution. A database 720 may be used to store data accessible by the system 700 by way of data interface 718.
FIG. 8 illustrates example metadata database management for a distributed file system 800 according to particular embodiments. The distributed file system 800 may use many of the same components that are illustrated in FIGS. 2A, 2B, and 3A-3H, and they may operate in a substantially similar manner. Therefore, in the interests of brevity and clarity, an explanation of the function and operation of these common components will not be repeated.
The maps 360 a and 360 c may be copies of a distributed data structure (e.g., distributed file share of storage items) that are maintained and accessed at each FSVM 170 a-c using a distributed data access coordinator 370 a and 370 c. The distributed data access coordinator 370 a and 370 c may be implemented based on distributed locks or other storage item access operations. Alternatively, the distributed data access coordinator 370 a and 370 c may be implemented by maintaining a master copy of the maps 360 a and 360 c at a leader node such as the host machine 200 c, and using distributed locks to access the master copy from each FSVM 170 a and 170 b. The distributed data access coordinator 370 a and 370 c may be implemented using distributed locking, leader election, or related features provided by a centralized coordination service for maintaining configuration information, naming, providing distributed synchronization, a metadata database for the directory or distributed file share, and/or providing group services (e.g., APACHE ZOOKEEPER or other distributed coordination software).
In particular embodiments, VFS 202 may enumerate contents of a directory or a distributed file share. In particular embodiments, distributed data access coordinator 370 a and 370 c may include metadata database engine 810 a and 810 c. Metadata database engine 810 a and 810 c may manage a metadata database that stores metadata information for some or all storage items in the directory or distributed file share. The metadata database may be stored at one of metadata database engine 810 a or 810 c (e.g., at a master one of the distributed data access coordinator 370 a and 370 c). The metadata stored in the metadata database may include a directory names, directory statistics, directory access control lists (ACLs), etc. When a request to enumerate contents of a directory or distributed file share is received (e.g., from a user through one or more user VMs 105 a-c and/or client 330 described herein), a distributed data access coordinator (e.g., one of the distributed data access coordinator 370 a or 370 c) may, responsive to the request, allocate an amount of memory space (e.g., buffer space) to store metadata pertaining to a number of directories (e.g., a metadata database). The distributed data access coordinator 370 a and 370 c may allocate a sufficiently large buffer for the metadata database in order to contain the metadata (including permissions information) for a large set of the storage items (e.g., 5K-10K or more directories) in the directory or distributed file share. After the allocation of the memory space, the metadata database engine 810 a and 810 c may work together to retrieve all metadata information associated the directory or distributed file share relevant to the request and update (e.g., add, remove, change, etc.) the information in the metadata database based on the retrieved information. Metadata may be fetched by the metadata database engine 810 a and 810 c for all storage items pertaining to the single request in a single query in order to enumerate the directory or distributed file share. In some examples, the VFS 202 may only include a single metadata database engine 810 a or 810 c in a master one of the distributed data access coordinators 370 a or 370 c to perform the management of the metadata database. Using one query to fetch metadata for multiple (e.g., thousands) of directories may avoid a less efficiency directory enumeration that involves multiple queries (e.g., per-directory queries). The metadata database may store some or all of the metadata in a pre-defined format, such as, by way of example and not limitation: |directory_1 name length|directory_1 name|directory_1 statistics size|directory_1 statistics|directory_1 ACL size|directory_1 ACL|dir_2 name length|directory_2 name|directory_2 statistics size| directory_2 statistics|directory_2 ACL size|directory_2 ACL| . . . |directory_n name length|directory_n name|directory_n statistics size|directory_n statistics|directory_n ACL size|directory_n ACL|. Other pre-defined formats may be used in other examples.
To evaluate whether a user has access to one or more directories, examples of distributed data access coordinators 370 a or 370 c may loop over the stored metadata in the buffer and check the ACL of each directory (e.g., one by one). This may avoid a need to utilize additional call(s) for finding the ACL of each directory again. The distributed data access coordinator 370 a or 370 c may compare the requestor's access controls with the ACL's listed to determine if a user has access to each directory. If the user has access, the directory may be included in a response to the enumeration request.
Once the metadata in the buffer has been evaluated, the distributed data access coordinator 370 a or 370 c may issue another call to the centralized database to obtain and store a next set of directory metadata and this procedure may be repeated until all entries from the given share have been evaluated. In this manner, multiple (e.g., thousands) of records may be obtained in one query from a centralized database, which may reduce the number of costly database calls. Moreover, techniques described herein may provide for faster enumeration of directories while evaluating whether a requestor has permission (e.g., ‘read’ access) to each directory.
Herein, a computer-readable non-transitory storage medium or media may include one or more semiconductor-based or other integrated circuits (ICs) (such, as for example, field-programmable gate arrays (FPGAs) or application-specific ICs (ASICs)), hard disk drives (HDDs), hybrid hard drives (HHDs), optical discs, optical disc drives (ODDs), magneto-optical discs, magneto-optical drives, floppy diskettes, floppy disk drives (FDDs), magnetic tapes, solid-state drives (SSDs), RAM-drives, SECURE DIGITAL cards or drives, any other suitable computer-readable non-transitory storage media, or any suitable combination of two or more of these, where appropriate. A computer-readable non-transitory storage medium may be volatile, non-volatile, or a combination of volatile and non-volatile, where appropriate.
Herein, “or” is inclusive and not exclusive, unless expressly indicated otherwise or indicated otherwise by context. Therefore, herein, “A or B” means “A, B, or both,” unless expressly indicated otherwise or indicated otherwise by context. Moreover, “and” is both joint and several, unless expressly indicated otherwise or indicated otherwise by context. Therefore, herein, “A and B” means “A and B, jointly or severally,” unless expressly indicated otherwise or indicated otherwise by context.
The scope of this disclosure encompasses all changes, substitutions, variations, alterations, and modifications to the example embodiments described or illustrated herein that a person having ordinary skill in the art would comprehend. The scope of this disclosure is not limited to the example embodiments described or illustrated herein. Moreover, although this disclosure describes and illustrates respective embodiments herein as including particular components, elements, feature, functions, operations, or steps, any of these embodiments may include any combination or permutation of any of the components, elements, features, functions, operations, or steps described or illustrated anywhere herein that a person having ordinary skill in the art would comprehend. Furthermore, reference in the appended claims to an apparatus or system or a component of an apparatus or system being adapted to, arranged to, capable of, configured to, enabled to, operable to, or operative to perform a particular function encompasses that apparatus, system, component, whether or not it or that particular function is activated, turned on, or unlocked, as long as that apparatus, system, or component is so adapted, arranged, capable, configured, enabled, operable, or operative.

Claims

1. (canceled)

2. A system comprising:

a plurality of host machines, each host machine configured to host at least one file server virtual machine (FSVM) to provide a plurality of FSVMs, the plurality of FSVMs configured to form a virtualized file server (VFS), the VFS including a file system hosted on each host machine;

each FSVM configured to present a namespace of storage items including a distributed file share, the distributed file share having a plurality of directories, the namespace implemented using a storage pool including local storage distributed across the plurality of host machines, each file system storing metadata associated with the plurality of directories;

wherein the VFS is configured to:

receive a request to enumerate the distributed file share of the namespace, the distributed file share managed across the plurality of FSVMs;

retrieve the metadata associated with the plurality of directories of the distributed file share; and

update a metadata database with the metadata.

3. The system of claim 2, wherein the plurality of host machines includes a first host machine configured to provide a first FSVM and wherein the first FSVM is configured to receive the request to enumerate the distributed file share.

4. The system of claim 2, wherein the metadata associated with the plurality of directories of the distributed file share is retrieved from each file system.

5. The system of claim 4, wherein each file system is associated with a respective FSVM hosted on the host machine.

6. The system of claim 2, wherein the file systems hosted on the host machines form a distributed file system.

7. The system of claim 2, wherein the VFS is configured to retrieve the metadata associated with the plurality of directories of the distributed file share using a single query.

8. The system of claim 2, wherein the VFS is further configured to allocate a buffer for the metadata.

9. The system of claim 8, wherein the VFS is further configured to iterate over entries in the buffer to build a response to the request to enumerate the distributed file share of the namespace.

10. At least one non-transitory computer readable media encoded with instructions, which, when executed, cause a virtualized file server (VFS) to perform actions, the actions including:

receive a request to enumerate a distributed file share of a namespace of storage items, the distributed file share having a plurality of directories, the distributed file share managed across a plurality of file server virtual machines (FSVMs), each of the plurality of FSVMs hosted by a respective host machine of a plurality of host machines, and wherein the plurality of FSVMs are configured to form the virtualized file server, the each of the plurality of FSVMs configured to present the namespace, the namespace implemented using a storage pool including local storage distributed across the plurality of host machines, the VFS including a file system hosted on each of the plurality of host machines, each file system storing metadata associated with the plurality of directories;

update a metadata database with the metadata.

11. The at least one non-transitory computer readable media of claim 10, wherein the plurality of host machines includes a first host machine configured to provide a first FSVM and wherein the first FSVM is configured to receive the request to enumerate the distributed file share.

12. The at least one non-transitory computer readable media of claim 10, wherein the actions comprise retrieve the metadata associated with the plurality of directories of the distributed file share from each file system.

13. The at least one non-transitory computer readable media of claim 12, wherein each file system is associated with a respective FSVM hosted on the host machine.

14. The at least one non-transitory computer readable media of claim 11, wherein the file systems hosted on the plurality of host machines form a distributed file system.

15. The at least one non-transitory computer readable media of claim 11, wherein the actions comprise retrieve the metadata associated with the plurality of directories of the distributed file share using a single query.

16. The at least one non-transitory computer readable media of claim 11, wherein the actions further comprise allocate a buffer for the metadata.

17. The at least one non-transitory computer readable media of claim 16, wherein the actions further comprise iterate over entries in the buffer to build a response to the request to enumerate the distributed file share of the namespace.

18. At method comprising:

receive, at a virtualized file server (VFS), a request to enumerate a distributed file share of a namespace of storage items, the distributed file share having a plurality of directories, the distributed file share managed across a plurality of file server virtual machines (FSVMs), each of the plurality of FSVMs hosted by a respective host machine of a plurality of host machines, and wherein the plurality of FSVMs are configured to form the virtualized file server, the each of the plurality of FSVMs configured to present the namespace, the namespace implemented using a storage pool including local storage distributed across the plurality of host machines, the VFS including a file system hosted on each of the plurality of host machines, each file system storing metadata associated with the plurality of directories;

update a metadata database with the metadata.

19. The method of claim 18, wherein the plurality of host machines includes a first host machine configured to provide a first FSVM and wherein the first FSVM is configured to receive the request to enumerate the distributed file share.

20. The method of claim 18, further comprising retrieve the metadata associated with the plurality of directories of the distributed file share from each file system.

21. The method of claim 20, wherein each file system is associated with a respective FSVM hosted on the host machine.

22. The method of claim 18, wherein the file systems hosted on the plurality of host machines form a distributed file system.

23. The method of claim 18, wherein the method further comprises retrieve the metadata associated with the plurality of directories of the distributed file share using a single query.

24. The method of claim 18, wherein the method further comprises allocate a buffer for the metadata.

25. The method of claim 24, wherein the method further comprises iterate over entries in the buffer to build a response to the request to enumerate the distributed file share of the namespace.