CN114003556A - Container deployment acceleration method based on delayed loading and P2P distribution - Google Patents
Container deployment acceleration method based on delayed loading and P2P distribution Download PDFInfo
- Publication number
- CN114003556A CN114003556A CN202111272621.1A CN202111272621A CN114003556A CN 114003556 A CN114003556 A CN 114003556A CN 202111272621 A CN202111272621 A CN 202111272621A CN 114003556 A CN114003556 A CN 114003556A
- Authority
- CN
- China
- Prior art keywords
- local
- mirror image
- tracker
- object storage
- tracking service
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/11—File system administration, e.g. details of archiving or snapshots
- G06F16/128—Details of file system snapshots on the file-level, e.g. snapshot creation, administration, deletion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/14—Details of searching files based on file metadata
- G06F16/148—File search processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/17—Details of further file system functions
- G06F16/172—Caching, prefetching or hoarding of files
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/18—File system types
- G06F16/182—Distributed file systems
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/104—Peer-to-peer [P2P] networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1095—Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Library & Information Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The invention realizes a container deployment acceleration method based on delayed loading and P2P distribution by a method in the field of distributed systems. By defining a new mirror image storage format to be stored in a local object storage pool utilizing an expandable distributed object storage architecture or a layered object position tracking architecture, the method realizes the removal of file redundancy among mirror images, reuses the existing data in scenes such as upgrading and the like, and realizes delayed loading and P2P distribution on the format; and on the basis of the mirror image storage format, designing a server side which is based on the mirror image storage format and consists of two modules, and finally returning corresponding data from a local object storage pool according to the name of the object. The method provided by the invention solves the problems of delayed loading of mirror image data, and file redundancy and non-reusability in mirror image storage, and simultaneously reduces the bottleneck of a mirror image warehouse in large-scale distribution by distributing the files in the mirror image through extensible distributed object storage.
Description
Technical Field
The invention relates to the technical field of networks, in particular to a container deployment acceleration method based on delayed loading and P2P distribution.
Background
The traditional container deployment mode has the problems of slow cold start, slow large-scale distribution, large storage redundancy and low data reuse degree. Cold boot refers to deploying a container on a node where the mirror does not yet exist, at which point the node needs to pull the mirror from the repository first and then boot the container. One study analyzed the commonly used mirror, indicating that the mirror pull process accounted for 76% of the cold start time of the container on average, while only 6.4% of the data was actually used. When a large number of nodes need to pull a mirror from the warehouse, the mirror warehouse is easily called a bottleneck of large-scale deployment. In addition, due to the layered packaging format of the image, a large number of files are redundantly stored. A study analyzed a large number of images on a Docker Hub, and found that only 3% of the files in the images were unique. However, in the current image format, redundant files between images cannot be reused, and existing data may be repeatedly downloaded when the image version is upgraded.
Existing container deployment acceleration methods can be divided into three categories:
the first type is a mirror store based on distributed storage, such as the Slacker, Cider, CernVM, and the like. The method stores the mirror image in a distributed file system in advance, realizes delayed loading through protocols such as NFS and the like, and realizes redundancy removal through the characteristics of the file system. The method has the disadvantages that the size of the mirror image warehouse is preset, waste is generated when the size is too large, and the bottleneck is easily formed when the size is too small.
The second type is by caching shared images, such as Wharf. The method enables a plurality of nodes to share a mirror image cache, and realizes delayed loading by using mirror image data in the cache through protocols such as NFS and the like. This is a compromise, and the process of caching the pull and unpack images from the image repository is consistent with the original image distribution mode. There is a limit in the effects of cold start acceleration and redundancy removal.
The third class is to define new mirror storage formats such as Stargz, DADI, D4C, etc. Where Stargz implements lazy loading via an addressable compressed packet format. The DADI reduces the mirror warehouse bottleneck by the block device ZFile format incorporating tree P2P distribution while delaying loading. D4C reorganizes the hierarchical order of the images so that data can be downloaded with pre-measured priority. This type of solution is flexible, possibly due to design priority issues, and their solutions do not simultaneously address the three aforementioned issues.
The above methods often only address one or two of the problems, and are incompatible with each other and cannot be simply combined. The method also belongs to the third category, and hopefully can simultaneously solve the three problems of slow cold start, data redundancy, difficulty in multiplexing and mirror image warehouse bottleneck.
Disclosure of Invention
Therefore, the invention firstly provides a container deployment acceleration method based on delayed loading and P2P distribution, which is characterized in that a new mirror image storage format is defined and stored in a local object storage pool utilizing an expandable distributed object storage architecture or a layered object position tracking architecture, so that the removal of file redundancy among mirror images is realized, the existing data is reused in scenes such as upgrading and the like, and the delayed loading and the P2P distribution are realized on the basis of the format;
designing local and remote structures based on the mirror image storage format;
the local end comprises a local mirror image management unit and a local object management unit, and the remote end comprises a remote mirror image warehouse and a remote object management unit;
the remote mirror image warehouse stores a mirror image layer compression package containing metadata files and mirror image description files; the remote object management unit comprises a remote object storage service end, a remote object storage client and an object position management unit;
the local object management unit comprises a local object storage client and a local object storage server;
the local mirror image management unit pulls data from the mirror image layer compressed packet and the mirror image description file through a daemon process and transmits the data to a snapshot management program, the snapshot management program sends mirror image layer metadata to a virtual file system, and the virtual file system reads data in a local object storage pool and sends an object loading request to a local object storage client;
the local object storage client receives an object loading request, writes an object into the local object storage pool, reports the object to an object position management unit for inquiry, and the remote object storage server acquires the object from the local object storage client;
and the remote object storage client side carries out report inquiry to an object position management unit, and the local object storage server side acquires the object from the remote object storage client side and reads the object from the local object storage pool.
The new mirror storage format is: the mirror image comprises configuration information and hierarchy information, the hierarchy information comprises file metadata and directory tree information, contents of directories and symbolic links are coded in the file metadata, and the file contents are found from the object storage by using a hash value of the file contents in the file metadata as a key.
The expandable distributed object storage architecture specifically comprises: an object position tracking service tracker arranged through fragmentation is respectively responsible for tracking a subset of objects, and a node is selected to acquire data when a node needs to acquire a certain object by setting a peer mirror image warehouse registry and a node host to be peer-to-peer;
the hierarchical object position tracking architecture divides the object position tracking service tracker into a Local object tracking service tracker and a Global object tracking service tracker on the basis of an expandable distributed object storage architecture, a node preferentially searches for an object in the Local object tracking service tracker, and the node searches for the Global object tracking service tracker only when a corresponding object cannot be found locally.
The specific method for tracking a subset of objects by the object position tracking service tracker is that when receiving a query from a client, only a subset of object positions is returned, so that the client has enough information to find a required object, when receiving a report from the client, the object position information needs to be updated, a heartbeat connection is maintained with the client, when the client cannot be connected with the client, the client is shielded from the position information, the object position tracking service tracker is also responsible for traffic management, a large number of nodes are prevented from simultaneously storing request objects to a small part of objects, and the subsequent client is selected to acquire the objects from more sources after waiting for the previous client to finish; the etcd is used for carrying out dynamic configuration management, when a node of an object position tracking service tracker fails, the load of the node is migrated to other nodes, and the nodes in the cluster periodically communicate with the etcd to acquire the latest configuration
The Local object tracking service Local tracker reports object position information to the Global object tracking service Global tracker in a periodic communication mode, when a certain client cannot find a certain object on the Local object tracking service Local tracker, the object position is inquired for the Global object tracking service Global tracker, and meanwhile, the Local object tracking service Local tracker records that the client acquires the object, so that other Local clients can be prevented from inquiring the Global object tracking service Global tracker at the same time.
The client is responsible for adding, deleting and reporting objects, when a new object needs to be obtained, a position list is obtained from an object position tracking service tracker, then the object is obtained from a target position according to a certain rule, and the change is recorded. Updating changes to an object location tracking service tracker in a periodic heartbeat message; the frequency of the fluctuating updates is inversely related to the number of copies of the object, requiring the object location to be updated faster when the number of copies of the object is small. When there are more copies of the object, location updates may be deferred to save on the number of communications.
The virtual file system is realized by using FUSE, a read-only file system is mounted according to metadata information of a mirror image layer, when a process requests to open a file, the virtual file system opens a corresponding file from a local object pool according to a file hash value, if the file does not exist in the local object pool, the object needs to be requested from an object loading module, and data reading expense brought by using FUSE is reduced by using FUSE passthrough.
The technical effects to be realized by the invention are as follows:
the invention provides a mirror image storage format, which supports delayed loading of mirror image data and solves the problems of file redundancy and non-reusability in mirror image storage; meanwhile, an object distribution mode is provided, and the bottleneck of a mirror warehouse in large-scale distribution is reduced by storing and distributing files in the mirror through an extensible distributed object.
Drawings
FIG. 1 is an overall architecture diagram;
FIG. 2 is a diagram of a new mirror format
FIG. 3 is a diagram of an extensible distributed object storage architecture
FIG. 4 a layered object position tracking architecture diagram
Detailed Description
The following is a preferred embodiment of the present invention and is further described with reference to the accompanying drawings, but the present invention is not limited to this embodiment.
The invention provides a container deployment acceleration method based on delayed loading and P2P distribution. The method firstly defines a new mirror image storage format, the storage format can remove the file redundancy among mirror images, and the existing data can be reused in scenes such as upgrading and the like; and implement lazy loading and P2P distribution on top of this format. Fig. 1 is an overall overview of the scheme, which consists of two main parts: the system comprises a local snapshot management module and a distributed object distribution module. The local snapshot management module is responsible for providing support for the new mirror image format and realizes delayed loading. The distributed object distribution module realizes the distribution of the mirror image actual data. The overall architecture of the scheme is shown in fig. 1, and the remote end in the figure can be a mirror repository or a peer node.
Local mirror image management module
The mirror format definition in the present invention is shown in fig. 2. The mirror image comprises configuration information and hierarchy information, wherein each layer of the mirror image only comprises file metadata and directory tree information, and does not comprise data blocks of files any more, so that the data volume required to be downloaded when the mirror image is pulled is greatly reduced. Where the contents of the directory and symbolic link are encoded in metadata, the hash value (sha256) of the file contents is stored in the metadata of the ordinary file, which can be used as a key to find the file contents from the object store.
Virtual file system:
using FUSE enables a virtual file system that can mount a read-only file system based on metadata information at the mirror level. When a process requests to open a file, it opens the corresponding file from the local object pool based on the file hash value. If the file does not already exist in the local object pool, the object needs to be requested from the object load module. The system uses FUSE passthrough to minimize the data reading overhead caused by using FUSE.
And (3) snapshot management program:
and realizing a snapshot management program for the file system, wherein in the process of image unpacking, the snapshot management program stores the unpacked metadata in a specific position. When requesting the layer, the snapshot manager mounts the layer for use by the container using the virtual file system described above. When creating a container, the snapshot management program creates an additional container read-write layer, and then combines all the mirror image layers into a unified file system through overlayfs to try to use the container. All modifications will occur on the container read-write layer in copy-on-write (copy on write).
Scalable distributed object store
The invention provides an extensible distributed object storage, which is similar to an HDFS in architecture and aims to reduce the bottleneck of a mirror image warehouse in large-scale container deployment. Fig. 3 depicts the architecture of the module. Wherein.
In larger scale scenarios (such as interpupillary), a hierarchy of object location tracking services may be organized, as shown in FIG. 4. The node can preferentially search for the object locally, and can search for the object in different places only when the corresponding object cannot be found locally.
Object location tracking service:
the object location tracking service (tracker) is responsible for the management of object locations. When a query is received from a client, it need only return a subset of the object locations so that the client has enough information to find the objects it needs. When a report is received from a client, it needs to update the object location information. It also maintains a heartbeat connection with the client, which masks the client from location information when it is not available to contact the previous client. The object location tracking service is also responsible for traffic management, preventing a large number of nodes from simultaneously requesting objects to a small portion of the object store. It may choose to have the following clients obtain objects from more sources after the preceding clients have finished.
In a layered architecture, the local tracker also needs to periodically communicate with the global tracker to report object location information. When the client cannot find an object on the local tracker, the client queries the global tracker for the object location. Meanwhile, the local tracker records that this client is acquiring the object, so as to avoid other local clients from querying the global tracker at the same time.
The object storage client:
the client is responsible for adding, deleting and reporting the objects. When a new object needs to be obtained, it obtains a position list from the object position tracking service, then obtains the object from the target position according to a certain rule, and records the change. In regular heartbeat messages it updates the object location tracking service with these changes.
The frequency of the fluctuating updates is inversely related to the number of copies of the object, requiring the object location to be updated faster when the number of copies of the object is small. When there are more copies of the object, location updates may be deferred to save on the number of communications.
The object storage server side:
the server side is responsible for responding to object requests from other clients and returning corresponding data from the local object storage pool according to the name (key) of the object. The services of the mirror repository may be provided collectively by a plurality of nodes, each of which may store a subset of all objects, and may store multiple copies for more popular objects.
Dynamic configuration management
To prevent the unavailability of services due to a single point of failure of the metadata service, dynamic configuration management is performed using etcd. When one object position tracking node fails, the load of the node can be automatically migrated to other nodes. The nodes in the cluster will communicate with the etcd periodically to get the latest configuration.
Claims (6)
1. A container deployment acceleration method based on delayed loading and P2P distribution, characterized in that: by defining a new mirror image storage format to be stored in a local object storage pool utilizing an expandable distributed object storage architecture or a layered object position tracking architecture, the method realizes the removal of file redundancy among mirror images, reuses the existing data in scenes such as upgrading and the like, and realizes delayed loading and P2P distribution on the format;
designing local and remote structures based on the mirror image storage format;
the local end comprises a local mirror image management unit and a local object management unit, and the remote end comprises a remote mirror image warehouse and a remote object management unit;
the remote mirror image warehouse stores a mirror image layer compression package containing metadata files and mirror image description files; the remote object management unit comprises a remote object storage service end, a remote object storage client and an object position management unit;
the local object management unit comprises a local object storage client and a local object storage server;
the local mirror image management unit pulls data from the mirror image layer compressed packet and the mirror image description file through a daemon process and transmits the data to a snapshot management program, the snapshot management program sends mirror image layer metadata to a virtual file system, and the virtual file system reads data in a local object storage pool and sends an object loading request to a local object storage client;
the local object storage client receives an object loading request, writes an object into the local object storage pool, reports the object to an object position management unit for inquiry, and the remote object storage server acquires the object from the local object storage client;
and the remote object storage client side carries out report inquiry to an object position management unit, and the local object storage server side acquires the object from the remote object storage client side and reads the object from the local object storage pool.
2. The container deployment acceleration method based on delayed loading and P2P dispatching of claim 1, wherein: the new mirror storage format is: the mirror image comprises configuration information and hierarchy information, the hierarchy information comprises file metadata and directory tree information, contents of directories and symbolic links are coded in the file metadata, and the file contents are found from the object storage by using a hash value of the file contents in the file metadata as a key.
3. The container deployment acceleration method based on delayed loading and P2P dispatching as claimed in claim 2, wherein: the expandable distributed object storage architecture specifically comprises: an object position tracking service tracker arranged through fragmentation is respectively responsible for tracking a subset of objects, and a node is selected to acquire data when a node needs to acquire a certain object by setting a peer mirror image warehouse registry and a node host to be peer-to-peer;
the hierarchical object position tracking architecture divides the object position tracking service tracker into a Local object tracking service tracker and a Global object tracking service tracker on the basis of an expandable distributed object storage architecture, a node preferentially searches for an object in the Local object tracking service tracker, and the node searches for the Global object tracking service tracker only when a corresponding object cannot be found locally.
4. The container deployment acceleration method based on delayed loading and P2P dispatching as claimed in claim 3, wherein: the specific method for tracking a subset of objects by the object location tracking service tracker is that when receiving a query from an object storage client, only a subset of object locations is returned, so that the object storage client has enough information to find a required object, when receiving a report from the object storage client, the object location information needs to be updated, and a heartbeat connection is maintained with the object storage client, when the last object storage client cannot be connected, the object storage client is shielded from the location information, the object location tracking service tracker is also responsible for traffic management, so that a large number of nodes are prevented from simultaneously requesting objects to store a small part of objects, and a subsequent object storage client is selected to acquire the objects from more sources after waiting for the completion of the previous object storage client; the method comprises the steps that dynamic configuration management is carried out by using the etcd, when a node of an object position tracking service tracker fails, the load of the node is transferred to other nodes, and the nodes in a cluster can communicate with the etcd regularly to obtain the latest configuration;
the Local object tracking service Local tracker reports object position information to the Global object tracking service Global tracker in a periodic communication mode, when a certain object storage client cannot find a certain object on the Local object tracking service Local tracker, the object position is inquired for the Global object tracking service Global tracker, and meanwhile, the Local object tracking service Local tracker records that the object storage client acquires the object, so that other Local clients can be prevented from inquiring the Global object tracking service Global tracker at the same time.
5. The container deployment acceleration method based on delayed loading and P2P dispatching of claim 4, wherein: the object storage client is responsible for adding, deleting and reporting objects, when a new object needs to be obtained, a position list is obtained from an object position tracking service tracker, then the object is obtained from a target position according to a certain rule, and the change is recorded. Updating changes to an object location tracking service tracker in a periodic heartbeat message; the frequency of the fluctuating updates is inversely related to the number of copies of the object, requiring the object location to be updated faster when the number of copies of the object is small. When there are more copies of the object, location updates may be deferred to save on the number of communications.
6. The container deployment acceleration method based on delayed loading and P2P dispatching as claimed in claim 5, wherein: the virtual file system is realized by using FUSE, a read-only file system is mounted according to metadata information of a mirror image layer, when a process requests to open a file, the virtual file system opens a corresponding file from a local object pool according to a file hash value, if the file does not exist in the local object pool, the object needs to be requested from an object loading module, and data reading expense brought by using FUSE is reduced by using FUSE passthrough.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111272621.1A CN114003556A (en) | 2021-10-29 | 2021-10-29 | Container deployment acceleration method based on delayed loading and P2P distribution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111272621.1A CN114003556A (en) | 2021-10-29 | 2021-10-29 | Container deployment acceleration method based on delayed loading and P2P distribution |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114003556A true CN114003556A (en) | 2022-02-01 |
Family
ID=79925271
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111272621.1A Pending CN114003556A (en) | 2021-10-29 | 2021-10-29 | Container deployment acceleration method based on delayed loading and P2P distribution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114003556A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115567547A (en) * | 2022-08-18 | 2023-01-03 | 网易（杭州）网络有限公司 | Mirror image distribution method, system and equipment |
| CN116048728A (en) * | 2023-01-16 | 2023-05-02 | 安超云软件有限公司 | Container mirror acceleration method based on-demand delay loading and application |
-
2021
- 2021-10-29 CN CN202111272621.1A patent/CN114003556A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115567547A (en) * | 2022-08-18 | 2023-01-03 | 网易（杭州）网络有限公司 | Mirror image distribution method, system and equipment |
| CN116048728A (en) * | 2023-01-16 | 2023-05-02 | 安超云软件有限公司 | Container mirror acceleration method based on-demand delay loading and application |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11755415B2 (en) | Variable data replication for storage implementing data backup | |
| CN111324360B (en) | Container mirror image construction method and system for edge calculation | |
| US10579610B2 (en) | Replicated database startup for common database storage | |
| US10437721B2 (en) | Efficient garbage collection for a log-structured data store | |
| US11687265B2 (en) | Transferring snapshot copy to object store with deduplication preservation and additional compression | |
| US6449731B1 (en) | Self-healing computer system storage | |
| TWI232382B (en) | A distributed storage system for data-sharing among client computers running different operating system types | |
| US8458181B2 (en) | Distributed free block map for a clustered redirect-on-write file system | |
| US9223843B1 (en) | Optimized log storage for asynchronous log updates | |
| CN102779185B (en) | High-availability distribution type full-text index method | |
| US7899850B2 (en) | Relational objects for the optimized management of fixed-content storage systems | |
| US8904006B2 (en) | In-flight block map for a clustered redirect-on-write filesystem | |
| US10725666B2 (en) | Memory-based on-demand data page generation | |
| US6530036B1 (en) | Self-healing computer system storage | |
| US20100161657A1 (en) | Metadata server and metadata management method | |
| US7870106B1 (en) | Client side caching in a global file system | |
| EP3811229B1 (en) | Hierarchical namespace service with distributed name resolution caching and synchronization | |
| JP2018049652A (en) | Log record management | |
| US20200285612A1 (en) | Defragmentation for objects within object store | |
| US20040030731A1 (en) | System and method for accessing files in a network | |
| CN114003556A (en) | Container deployment acceleration method based on delayed loading and P2P distribution | |
| CA2574735A1 (en) | Metadata management for fixed content distributed data storage | |
| CN111190547B (en) | Distributed container mirror image storage and distribution system and method | |
| EP2619695A2 (en) | System and method for managing integrity in a distributed database | |
| US20200285611A1 (en) | Metadata attachment to storage objects within object store |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |