CN1645342A - Large scale resource memory managing method based on network under SAN environment - Google Patents
Large scale resource memory managing method based on network under SAN environment Download PDFInfo
- Publication number
- CN1645342A CN1645342A CNA2005100112313A CN200510011231A CN1645342A CN 1645342 A CN1645342 A CN 1645342A CN A2005100112313 A CNA2005100112313 A CN A2005100112313A CN 200510011231 A CN200510011231 A CN 200510011231A CN 1645342 A CN1645342 A CN 1645342A
- Authority
- CN
- China
- Prior art keywords
- command
- disk
- physics
- space
- metadata
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000008569 process Effects 0.000 claims abstract description 35
- 238000012545 processing Methods 0.000 claims description 47
- 230000006870 function Effects 0.000 claims description 43
- 239000013307 optical fiber Substances 0.000 claims description 24
- 230000015572 biosynthetic process Effects 0.000 claims description 14
- 238000012217 deletion Methods 0.000 claims description 14
- 230000037430 deletion Effects 0.000 claims description 14
- 230000004044 response Effects 0.000 claims description 14
- 230000008859 change Effects 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 9
- RGNPBRKPHBKNKX-UHFFFAOYSA-N hexaflumuron Chemical compound C1=C(Cl)C(OC(F)(F)C(F)F)=C(Cl)C=C1NC(=O)NC(=O)C1=C(F)C=CC=C1F RGNPBRKPHBKNKX-UHFFFAOYSA-N 0.000 claims description 7
- 230000007246 mechanism Effects 0.000 claims description 7
- 210000001519 tissue Anatomy 0.000 claims description 7
- 238000013467 fragmentation Methods 0.000 claims description 6
- 238000006062 fragmentation reaction Methods 0.000 claims description 6
- 239000007858 starting material Substances 0.000 claims description 5
- 210000002808 connective tissue Anatomy 0.000 claims description 3
- 238000005194 fractionation Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 3
- 230000008707 rearrangement Effects 0.000 claims description 3
- 238000007726 management method Methods 0.000 abstract description 23
- 238000004458 analytical method Methods 0.000 abstract description 2
- 230000000875 corresponding effect Effects 0.000 description 28
- 230000006855 networking Effects 0.000 description 15
- 238000012360 testing method Methods 0.000 description 7
- 230000008520 organization Effects 0.000 description 6
- 238000013468 resource allocation Methods 0.000 description 6
- 230000003287 optical effect Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000004321 preservation Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 230000001934 delay Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 239000012467 final product Substances 0.000 description 2
- 230000000873 masking effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011056 performance test Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 230000033772 system development Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A management method of large scale storage resource includes maintaining a set of storage resouce disposal information by node machine of special processor to manage different storage device resource in storage network so as to provide virtual storage service for front end host by carrying out process of command analysis with interface middle layer of small computer interface. The node machine with information list and software target machine can convert virtual access to be access of physical device.
Description
Technical field
Based on network large scale resource memory managing method belongs to field of storage area networks in the SAN environment, relates in particular to technical field of virtualization wherein.
Background technology
Along with fast development of information technology, people's data volume to be processed constantly increases, and has caused the great demand to storage system capacity and performance, and makes traditional storage mode can't satisfy user's needs.Generally speaking, every main frame all has the memory device of oneself, and this makes information dispersion on each main frame, forms information island, is unfavorable for information integrated.And the network storage can provide the Sharing Information access service, and has vast capacity and outstanding performance, therefore becomes the trend of storage system development.(Storage Area Network SAN) is a kind of special storage networking to storage area network, and it can link together various memory devices, form the storage resources of unified management, and main frame can visit these resources by optical fiber or IP network.The SAN system has the extensibility of good performance and height.In this storage architecture, the configuration of resource is constantly to change, and new equipment can constantly join in the system, to satisfy the demand of user to memory capacity or performance.
But, how to distribute every used equipment of main frame to become a serious problem along with the continuous increase of storage networking scale.At this moment, each main frame of front end may use various operating system, and the memory device in the network also has different specifications and type.Storage Virtualization work is exactly in order to address these problems.Traditional Intel Virtualization Technology has two kinds of main solutions, and the firstth, on each main frame, realize by agent software, the more common LVM that comprises in the linux system, this method is not suitable for multi-machine surroundings, is not suitable for different operating system yet; Second kind of scheme is to realize at the memory device end, and the array apparatus of a lot of manufacturers also provides virtualization, but this method can not be managed polytype equipment generally all towards particular device.These two kinds of schemes all can't satisfying magnanimity isomery storage networking environment.
In order to solve the resource management problem in the magnanimity isomery storage networking environment, we have proposed based on network large scale resource memory managing method in the SAN environment.It realizes the Storage Virtualization function on a special processing node in storage networking, and has following characteristics:
1. Gao Du compatibility: irrelevant fully with the operating system of front end main frame, and can manage various types of memory devices.
2. succinct resource information organizational form: only need on the node machine, take the storing device information that the magnanimity scale is preserved in a spot of space.
3. Gao Du dirigibility: because this method realizes that level is lower, thus can select the equipment that performance is good in the network preferentially to use neatly, thus equipment load balance guaranteed, improve the overall performance of storage system.
Summary of the invention
The objective of the invention is to realize that a kind of virtual method manages various types of memory devices in the mass memory network, and virtual memory space is offered the front end main frame use.This invention is primarily aimed at the demand of mass memory network environment, has designed and Implemented the practicable Storage Resource Management (SRM) method of a cover.By this method, can effectively reduce the resource requirement of facility information, and the allocation of space mode to user transparent is provided, reduce the complexity of resource management, improve the overall performance of storage system simultaneously.
Core of the present invention is to safeguard a cover storage resource configuration information by dedicated processes node machine, manage various memory device resource in the storage networking concentratedly, and carry out order analysis by SCSI (small computer system interface Small Computer SystemsInterface) middle layer and handle, thereby provide transparent virtual stores service fully to the front end main frame.Concrete realization framework comprises following two major parts:
1. the configuration information of the whole memory devices of node on-board maintenance (comprising physical equipment and virtual unit) tabulation: each memory device in the network is produced a structure, write down its various attributes and available space; After the keeper creates virtual disk, also each virtual disk is created a structure, write down various attributes and logical space allocation scheme.
2. realize the SCSI software object device STML (SCSI Target Mid-levelLayer) of a virtual support function on the node machine, its work is that the front end main frame is converted into visit to actual physical device to the visit of virtual disk logical space, and the data processed result of physical equipment is fed back to the front end main frame.Logical order carries out for the basis according to storage resources organizational information above-mentioned to the conversion of physics order.
The present invention is a kind of based on network (realizing on network node VIC), and the Storage Virtualization solution that realizes on data path is called the AXUM system.This implementation method makes that its compatibility is very strong, is not subject to different host operating systems and device type fully.Simultaneously, we have determined unique resource information organizational form by the investigation to storage resources information in the mass memory environment.This organizational form not only allocation of space is flexible, and it is little to take up room, and is applicable to the Storage Resource Management (SRM) of magnanimity scale.Secondly, by the evaluation to different storage device, good equipment comes the assignment logic space can preferentially to select performance automatically, effectively reduces management difficulty.In addition, by the control of STML to each command process, we can add various optimized Measures easily, improve the entire system performance.
The operation steps of based on network large scale resource memory managing method is roughly (see figure 1):
1. load-on module:
All memory devices in magnanimity resources configuration management module (axum.o) probe network, each physical equipment that will manage abstract be a Physical Disk (PD) structure; From the metadata dish, obtain the virtual disk information that has disposed simultaneously, each virtual disk abstract be a Virtual Disk (VD) structure, organize the space layout of VD, and adjust the space operating position of corresponding PD according to the physical space layout of VD;
SCSI software object device STML module (mid_target.o) is organized the scsi device queue according to the configuration information of axum.o, sets up the command process thread simultaneously, is used for handling the I/O request of sending.
2. resource distribution: the keeper can carry out virtual configuration to the mass memory resource by the configuration tool that the node machine provides, as interpolation, the removal of PD, and the establishment of VD, deletion, dilatation and main frame appointment etc.After the user carried out resource distribution, the axum.o module was revised corresponding information organization structure and content in kernel.
3. metadata is preserved: system is at the complete storage resource configuration information of node on-board maintenance one cover, if the user makes amendment to resource distribution, cause configuration information to change, need so content corresponding is saved in the metadata dish, so that can recover original configuration when reloading in system.
4. main frame uses: the front end main frame drives by loading sonet card, the virtual disk that can find and use the node machine to provide, and actual physical equipment is transparent to the front end main frame.
5. data access: the front end main frame is converted into scsi command and sends to the node machine by optical-fibre channel the visit in virtual disk space.In the mid_target.o of node machine module, spatial organization's information according to axum.o is carried out address translation, visit to logical space is converted into visit to actual physical device, and the data of access are split or merge according to the organizational form of logical space, at last actual physics process of commands result is returned the front end main frame by the optical fiber drive feedback.
The present invention is based on network equipment realization, and except the management software interface, all other program all realizes on node machine VIC.The software architecture of AXUM virtualization system as shown in Figure 2.
The invention is characterized in: it is at following storage area network, promptly realizes among the SAN, and described network contains:
Main frame, promptly front-end server has more than 1, and wherein one as management host, the installation administration program interface;
In described disk, two metadata disks are arranged, be used to preserve all metadata informations, described metadata is meant in user's use, disc information after the each change that forms during the operation of the establishment that comprises physical disk, virtual deposit that takes place, deletion, expansion, described metadata comprises following content:
The basic metadata of PD position comprises that the overall situation of stored physical equipment is unified identifier, represent with UUID, and type information is interior;
The basic metadata of SC, SC is meant the storage and uniform space that various memory devices are formed, the basic metadata of SC is the data of SC message structure in the system;
The metadata of all PD of SC, i.e. all actual memory devices among the SC, i.e. the UUID information of all PD;
The metadata of the VD that SC is all comprises not only that promptly all VD are the data of virtual disk message structure, also comprise each VD in actual disk shared position, size in interior information;
Except that the metadata dish, all the other disks of system are only preserved the basic metadata of PD separately;
When method of the present invention was implemented in above-mentioned storage area network, it contained following steps successively:
The 1st step: at described I/O node machine is to load in the VIC node with lower module:
The resource distribution module is represented with axum.o; Its definite virtual disk parameter that will create, comprise the title, size of virtual disk, whether band tissue and data fragmentation number, interior nuclear symbol, kernel information interface are provided simultaneously and register special character type equipment, so that described module can receive user's virtual configuration-direct, and revise the respective resources configuration information that it is safeguarded, simultaneously the information that changes is saved in the metadata dish;
Scsi target device module is represented with mid_target.o, and it organizes the scsi device formation according to the configuration information of axum.o, sets up the command process thread simultaneously, and the I/O that sends in order to processing asks; For this reason, it sets up the scsi command that the thread of dummy command and physics order handles at virtual disk respectively is dummy command, and the scsi command for actual physical disks after dummy command decomposed is the chain phenotype formation of physics order;
When the above-mentioned I/O of processing asks, carry out the fractionation or the merging processing of map addresses and data according to the configuration information of axum.o, return result;
The 2nd step: resource distribution; The keeper carries out virtual configuration by the configuration tool that the node machine provides to the mass memory resource, and it contains following steps successively:
The 2.1st step: the user is by the definite virtual disk parameter that will create of resources configuration management software;
The 2.2nd step: the axum.o module finds that by ioctl mechanism the notice kernel that user's space transmits carries out the command id that virtual disk is created, the parameter that will pass to user's space kernel is taken out, comprise that virtual disk is title, size, organizational form and the strip data burst number of VD, carry out virtual disk and create;
The 2.3rd step: creating the virtual disk structure is the variable of ax_vd_t, inserts parameter information, and writes down new virtual disk configuration data;
The 2.4th step: judge the organizational form in the parameter, carry out corresponding allocation of space and handle:
The 2.4.1 step: when being connective tissue's mode:
Determine at first whether the virtual disk in the parameter is the new disk of creating, if not, then preferentially select the affiliated physical equipment of final stage physical space of this disk to carry out spatial spread; Otherwise, preferentially use best physical disk to carry out allocation of space;
Secondly, determine the space size that residue will be distributed, select new disk to carry out allocation of space;
The allocation of space success is returned;
The 2.4.2 step: when being the band organizational form, suppose that band quantity is s:
At first, determine whether to be new virtual disk, if then, select s physical disk to carry out allocation of space to all physical equipment orderings; Otherwise, select in original virtual disk the last pairing physical disk of s piece storage space to carry out allocation of space;
The space size that definite residue will be distributed;
Upgrade the capacity of virtual disk and the size of remaining space;
To the physical equipment rearrangement, select s new equipment;
The allocation of space success is returned;
The 2.5th step: fill new virtual disk structure, and this structure is inserted in the global variable, notify user's virtual disk to create and finish;
The 3rd step: metadata is preserved, and it contains following steps successively:
When operating, at first write the basic metadata of PD in the basic metadata of operated physical disk PD position at physical disk;
The position of preserving the basic metadata of SC in the metadata dish writes the information data of SC;
The position of preserving all VD metadata of SC in the metadata dish writes the basic metadata of VD and the occupied information of VD;
The 3rd step: data access
The 3.1st step: mid_target.o receives the SCSI read write command to the virtual disk logical space that front-end server is sent through interface function, and generates the corresponding instruction structure, coomand mode is set after, these imperative structures are inserted the processing queue of dummy command;
The 3.2nd step: mid_target.o wakes the dummy command thread up, calls Virtual Analysis instruction, judges and at the different operation of different command execution;
The 3.3rd step: read command is handled, and comprises following steps successively:
According to the virtual configuration information that the axum.o module is safeguarded, the mapping function of using by resource distribution is mapped to several physics scsi commands to dummy command;
To each physics order, make up the corresponding instruction structure, simultaneously, the state of dummy command and the state of each physics order are set;
The formation of physics command process is inserted in these physics orders, wake physics command process thread up and handle, and the respective handling state of dummy command is set;
After the physics command process finishes, the dummy command processing threads is collected the response of all physics orders, forms the response of dummy command, and the data that each physics order is returned merge simultaneously, then the response of this dummy command is sent to the optical fiber driver card, and revise a coomand mode;
The dummy command processing threads is deleted the corresponding physical order from respective queue, this dummy command is deleted from the dummy command processing queue again;
The 3.4th step: write order is handled, and comprises following steps successively:
According to the virtual configuration information that the axum.o module is safeguarded, the mapping function of using by resource distribution is mapped to several physics orders to dummy command;
For each physics order, make up the corresponding instruction structure, simultaneously, the state of dummy command and the state of each physics order are set;
The formation of physics command process is inserted in these physics orders;
The dummy command processing threads drives from the front-end server forward data according to the state calling interface function notice sonet card of dummy command, resets the state of dummy command and corresponding each physics order;
Sonet card drives and gives the dummy command processing queue data by interface function, and dummy command and each physics order are set as the state of handling;
The dummy command processing threads data fragmentation and be dealt in the corresponding physics order, is reset the state of dummy command and each physics order by interface function;
Physics command process thread is handled;
After disposing, drive notice front end main frame by sonet card, and the state of each physics order of change dummy command and correspondence, the dummy command processing threads is deleted the corresponding physical order from respective queue, this dummy command is deleted from the dummy command processing queue again.
The present invention is in order to manage independently the various storage resources in the mass memory network, reduces administration overhead and designs and develop.It is realized by add special processing node in storage networking.Adopt the present invention, as long as in storage networking, add a processing node machine, just can manage various types of memory devices easily, and provide service, therefore can adapt to the operating platform and the memory device of various isomeries in the mass storage system (MSS) fully the front end main frame of various platforms.The present invention provides perfect Storage Resource Management (SRM) service by unique resource information organizational form, choice of equipment mode and order control mode, and improves the readwrite performance of system.Simultaneously, the information space demand of maintenance required for the present invention is little, and allocation of space is reasonable flexibly, is applicable to the mass memory network environment.The present invention has realized in the mass memory network TH-MSNS that is researched and developed voluntarily by Tsing-Hua University and test.And provide virtual disk to carry out file access and various application service to the front end main frame (Windows, Linux etc.) of different operating platform on this basis.
Description of drawings
Fig. 1. large scale resource memory managing and command process flow process.
Fig. 2. the software module framework on the network processing node VIC.
Fig. 3. the hardware configuration of mass memory network and tissue.
Fig. 4. the resource information institutional framework of mass memory network.
Fig. 5. the logical space of virtual disk distributes.
Fig. 6. create the program circuit of virtual disk work.
Fig. 7. the organizational form when metadata is preserved.
Fig. 8. the scsi target device middle layer STML of virtual support function.
Fig. 9. during read-write requests, the state exchange of logical order and physics order.
The virtual disk readwrite performance that Figure 10 .VIC node provides.
The software delays that Figure 11 .VIC virtualization process is brought.
The data-handling capacity of Figure 12 .VIC node.
Embodiment
1. the configuration of hardware device and tissue
We are this cover independent development, and towards the virtualization system called after AXUM of mass memory network, and the processor of operation AXUM system is called the VIC node in the network.As the storage networking institutional framework of AXUM running environment as shown in Figure 3.All hardware devices comprise front end main frame, VIC node, common I/O processing node and optical fiber disk unit, all are connected on one or more optical fiber switch by optical fiber cable.The front end main frame has one or more, and the optical channel card (FC-HBA) of starter pattern is installed, and is used for queried access and the available memory device of visit fiber optic network; Wherein a front end main frame needs the installation administration interface as management host.Common I/O processing node is equipped with the optical channel card of target mode, below the disk or the disk array of connection ID E or SCSI type, make them can be activated the optical channel card visit of device pattern.The VIC node is the management node that adds large scale resource memory managing module (axum.o) and SCSI software object device module (mid_target.o), and exchanges with management host by the management software Agent.Be inserted with two FC-HBA cards on the VIC, one is used for the starter pattern, and another piece is used for target mode.VIC can be as independent processing node, also can own carry IDE or scsi disk equipment as network storage resource.
Utilize the sectoring function of switch that whole storage networking environment is divided into two district: Zone 1 and comprise all front-end servers and VIC, VIC is by the request of object machine FC-HBA response front end main frame, and the front end main frame can only be seen the equipment that VIC provides; Zone 2 comprises VIC and other all rear end memory devices, comprises FC disk unit and scsi disk array etc.VIC utilizes starter FC-HBA, can detect all equipment and carry out unified management.Because whole storage networking is divided into two districts, the main frame of front end will can not be found the physical equipment in the network, and can only detect the virtual disk that VIC offers front end.Simultaneously, utilize LUN masking (LUN masking) function, make different main frames can find and visit different virtual disks.
2. the information organization structure of mass memory resource
The AXUM system is organized into a unified storage pool to all amount of physical memory (comprising physical disk and LU equipment).The keeper can divide allocation of space and give virtual disk according to user's needs from storage pool, and offers different front end main frame uses.The resource allocation information institutional framework of AXUM as shown in Figure 4, wherein the main information structure comprises:
SC: the storage and uniform space that various memory devices are formed.
PD: rear end actual physical memory device, information such as recording capacity size and device identification.
FS: every section unallocated storage space reference position and size of giving virtual disk among the record PD.
VD: offer the virtual disk that front end uses, comprise capacity, information such as spatial mappings mode.
The minimum space unit of FS and PS (being called a SG) variable size wherein, the SG default size is 64KB.
In the information organization structure of AXUM, the size of each PD structure is about 100 bytes, and its FS quantity what the quantity of information of a PD depends primarily on, and the size of one of them FS structure is 12 bytes.Just can on PD, produce one section new unallocated space owing to have only when certain VD space of deletion, so the project in the FS chained list of corresponding PD can be not a lot.Suppose that each equipment has 10 sections unassigned zones, so only needs the information space about the 100+10*12=220 byte.Even manage up to ten thousand disks like this, also only can take VIC system resource (10000*220B=2.2MB) seldom.Similarly, what of its PS quantity are the quantity of information of VD depend primarily on.The size of VD structure is about 300 bytes, and the size of each PS structure is 16 bytes.In the mass memory network environment, the user may create a lot of VD, and each VD also may comprise many PS.Suppose to have created in system 1000 virtual disks, each VD size all is 100GB.Even each VD comprises 1000 PS like this, the metadata of all VD also only need take the storage space about 16MB.Therefore, AXUM preserves the space requirement altogether of mass memory resource allocation information less than 20MB.
Relatively, if we use the most frequently used LVM system to carry out virtual configuration, top environment needs the space of 180MB to be used for the preservation of resource allocation information at least.Simultaneously, LVM need will preserve information on every front-end server, and as long as the AXUM system is at VIC node on-board maintenance one cover metadata information.As seen, the information organization mode of AXUM is more suitable in the mass memory environment.
3. system loads
VIC node machine is exactly the logical server of a Daepori in fact, operation (SuSE) Linux OS (the above kernel of 2.4.18), two parts of AXUM virtualization system (configuration information structure and STML) are made kernel module respectively and are loaded, and also will start the Agent of management software simultaneously:
1) loads resource distribution module (axum.o).After this module loading, will finish following step successively:
Variate-value carries out initialization, and an interior nuclear symbol axum_SC is provided, and record resource distribution situation makes STML can utilize its inquiry corresponding information;
Provide "/proc " kernel information interface under the catalogue, so that user or management software are inquired about existing resource allocation information;
This module receives user's virtual configuration-direct, and revises the respective resources configuration information that it is safeguarded, simultaneously the information that changes is saved in the metadata dish.
2) load STML module (mid_target.o).After this module loading, will finish following step successively:
Automatically the virtual disk (VD) in the search system, and the virtual disk that searches formed a chained list with the form of stml_device, use for front end.
This module receives scsi command, carries out the fractionation or the merging processing of address translation and data, the result that last return command is handled according to the configuration information of axum.o.
3) the management software Agent on the VIC node machine:
Working procedure sagent produces and acts on behalf of finger daemon, and is formulating spanned file info and fmid under the catalogue automatically, respectively the intermediate result of hold-over command execution result and lunmask configuration.
The agency is undertaken alternately by the management software on Socket mechanism and the management host, the order that receiving management software is sent, and the corresponding function that calls the axum module is carried out operations such as the establishment of establishment, deletion, dilatation and the PD of VD, deletion, and returns execution result.
The information interface of agency by the VIC node, promptly file/proc/axum/pd ,/proc/axum/vd ,/proc/axum/sc, can obtain virtual configuration relevant information, and feed back to management software and show.
According to the order of management software, the agency can realize the dynamic-configuration to the lunmask file, and then realizes the dynamic-configuration to front end main frame and virtual disk.
We as management host, are configured a main frame of storage networking front end by the mass memory resource of administration interface program to the VIC management:
Communicate by letter with the VIC node-agent by Socket mechanism, obtain relevant information and show.
The user can obtain the relevant information of safeguarding on the VIC node about physical equipment, virtual unit, whole storage pool space statistical conditions and front end main frame by management software.
4. resource distribution
After each module loading of AXUM system, if never carried out virtual configuration, the front end main frame can not seen any virtual disk equipment so.Therefore, we must utilize the existing device space to create virtual disk by management software on VIC.The AXUM virtualization system need be in the enterprising row labels of physical equipment of its management, and reservation disk space, so, in order not influence the special use configuration of user to some disk unit, we need will allow each the equipment of AXUM management add virtualization system one by one to by the user, and virtual disk can only specify the device space that joins in the system to constitute by the user.Just do not carry out virtual configuration if add any physical equipment, system can provide the error message prompting.
The resource information institutional framework of AXUM makes that the distribution of logical space is very convenient.When increasing logical space, if the burst number of this virtual disk VD (stripe number) is S (S 〉=1), AXUM needs at every turn to divide the space from S FS that adheres to different PD separately so, and corresponding S PS section offered VD as the virtual disk logical space.Similarly, if reduce the size of virtual disk, need delete S (or integral multiple of S) section space last part from the PS table of VD, and be returned to the FS table of corresponding PD.After front-end server detects the virtual disk capacity and changes, just can be easily the capacity of this VD be carried out online modification.With Fig. 5 is example, and the VD among the figure is the virtual disk of a band tissue, and the burst number is 5.If increase the capacity of 50GB, the PS chained list rear end that the only space that need obtain 10GB respectively from 5 physical equipments, and 5 sections corresponding PS joins VD gets final product.And if deletion 20GB space, as long as in the end among 5 PS, every section space size that all reduces 4GB gets final product.
What common virtualization system was managed all is more small-scale storage resources, generally can't consider the difference of actual physical device, is easy to cause the waste and the performance decrease in space like this.The AXUM system towards be the mass memory environment, can supervising the network in all types of equipment of each manufacturer.Because the performance index of distinct device have nothing in common with each other, therefore we have added appraisal of equipment mechanism in AXUM, it is by the intensity of load of statistical physics equipment, space utilization situation and performance index etc., obtain the evaluation of estimate of each equipment, preferentially utilize the best equipment of those performances to carry out the distribution of logical space,, and guarantee the load balancing of each equipment as far as possible so that improve the readwrite performance of whole storage networking.
When carrying out appraisal of equipment, the main factor of considering comprises:
1) priority facility: specified by the user, virtual disk can preferentially use the high equipment of priority like this.
2) the read write command intensity Drw of PD: it equals the read-write requests total amount of this PD processing and the ratio of this PD creation-time.
3) allocatable space of PD size FSt: the space size that this physical equipment is still unallocated gives virtual disk.
4) the VD quantity Nvd on the PD: the VD sum that uses this physical equipment space.
5) the hardware index Hs of PD: the performance of reflection physical equipment comprises the space size, seek time, data transmission rate and equipment cache size etc.
The AXUM system to the evaluation function of each PD is: Hs*Fst+a/ (Nvd+1)+b/Drw, a wherein, b is a constant, can according to different indexs stress adjust.Under the identical condition of each PD priority, system at first chooses the allocation of space that the highest PD of evaluation of estimate carries out virtual disk.To each PD, from its FS chained list, divide the space and give VD.If desired volume less than FSt, is divided the space from the only FS section of size so at every turn.
The virtual configuration feature of the storage resources of AXUM comprises establishment, deletion, the capacity modification of virtual disk, and the use main frame appointment of virtual disk etc.Be how resource allocation information in the example illustrative system changes (Fig. 6) to create virtual disk below:
Whether 1) user determines the virtual disk parameter that will create by resources configuration management software, comprises the title of virtual disk, size, band tissue and data fragmentation number.Management software is filled into these information among the variable avd of a virtual disk structure (ax_vd_t) type.Be delivered to kernel spacing by ioctl mechanism from the user's space of (SuSE) Linux OS: ioctl (dev_fd, AXIO_VD_CREATE ， ﹠amp; Avd).Wherein dev_fd is the device number of the character type equipment that defines in kernel of AXUM resource distribution module, and AXIO_VD_CREATE is the command id that the notice kernel carries out the virtual disk building work, and avd is the parameter that user's space will pass to kernel.
2) after the axum.o module is ordered by the machine-processed AXIO_VD_CREATE that finds that user's space transmits of ioctl, parameter information in the avd address is obtained out, call virtual disk and create function: do_vdcreate (avd.vd_name, avd.vd_size, avd.mapping_type avd.stripe_num) carries out virtual disk and creates, and wherein four parameters are respectively the title of virtual disk VD, size, organizational form (supporting continuation mode and band mode at present) and strip data burst number.
3) the do_vd_create function is at first created the variable thevd of an ax_vd_t structure, is used for writing down new virtual disk configuration information, and parameter information is inserted, and judges the organizational form in the parameter then, carries out different allocation of space respectively and handles:
3.1) if connective tissue's mode, call continuous space partition function alloc_vdspace_linear (thevd, vdsize)
Carry out the virtual disk allocation of space, wherein first parameter is new virtual disk structure, and second parameter is the virtual disk size:
Determine the space size sg_remain that will distribute, when sg_remain greater than 0 the time, following steps are done in circulation:
The space that pfs is pointed to is allocated to pps as far as possible, and pps is inserted last that PS tabulates among the thevd.
Upgrade the value of virtual disk size and sg_remain, upgrade the information of corresponding PD.
The allocation of space success is returned.
3.2) if the band organizational form, call band allocation of space function alloc_vdspace_stripe (thevd, vdsize)
Carry out the virtual disk allocation of space, wherein first parameter is new virtual disk structure, and second parameter is the virtual disk size:
● determine in the equipment of each ppd sensing minimum allocatable space size min;
● from the equipment that each ppd points to, divide the min space and give virtual disk;
Upgrade the capability value of virtual disk structure thevd and remain the space size sg_remain that will distribute.
To the PD rearrangement, select s new equipment.
The allocation of space success is returned.
4) fill other statistical informations in new virtual disk (thevd) structure, comprise space size and virtual disk state etc., and this structure is inserted in the axum_SC global variable.Notify user's virtual disk to create successfully.
5. metadata is preserved
The user is in carrying out mass memory resource distribution process, the establishment of generation such as physical disk, deletion, during operation such as the establishment of virtual disk, deletion and expansion, the pairing information about physical storage device and virtual disk of the global variable axum_SC in the system will be changed.If these information are not preserved, will lose after the system restart.Therefore each change all must be kept at the information after the change on the disk.These be kept on the disk information data we be referred to as metadata.
The metadata of preserving comprises four types: the basic metadata of (1) PD, and it comprises that the overall situation of stored physical equipment unifies information such as identifier (UUID) and disk type; (2) the basic metadata of SC, the i.e. data of SC message structure in the system; (3) metadata of PD, the i.e. information of the UUID of all PD among the SC under the SC; (4) metadata of VD under the SC, it not only comprises the data of all VD message structures, also comprises the data of each VD information such as shared position, size in actual disk.
We vacate the space of preceding 25M and are used to deposit metadata in system on each actual physical disk.The layout of metadata in disk as shown in Figure 7.
In the system of reality, we use two disks to preserve all metadata informations, and we are referred to as the metadata disk, and remaining disk is only preserved the basic metadata of PD separately.This is because if all disks are all preserved a complete metadata, and the derivation expense of each metadata can be very big.Use two disks to preserve all metadata and neither have very big expense, and also have another one metadata disk to use after certain metadata disk breaks down, system is recovery configuring still, continues to use.
When the user carries out establishment, the deletion of physical disk, during operation such as the establishment of virtual disk, deletion and expansion, system can call the preservation operation of a metadata automatically.The process that system carries out metadata preservation operation is as follows:
6. data access
The front end main frame can detect the virtual disk that the AXUM system provides after loading the driver of optical channel card (FC-HBA).These virtual disks can be taken as common scsi disk and operate, and such as format, create file system, carry (mount) etc., and the data access visit just can be carried out to this virtual disk in initialization success back.The front end main frame sends by scsi command the read-write of virtual disk logical space, and scsi command can be sealed dress by FC, and the STML module (mid_target.o) that sends to the VIC node is handled.The treatment scheme of virtual support function STML as shown in Figure 8.Object machine FC-HBA drives the scsi command that comprises logical address is submitted to STML, and STML is organized into formation to these logical orders.For each logical order, if the order of inquiry class produces feedback information with the direct information content and SCSI standard format according to AXUM.Main querying command and disposal route thereof are:
| Order | Handle |
| REPORT?LUNS | Return Logical Disk quantity and determine information frame length |
| INQUIRY | Return the title of Logical Disk, information such as model |
| TEST?UNIT?READY | Command processing result information is changed to DID_OK |
| READ?CAPACITY | Return the data block quantity and the data block size of counterlogic dish |
Because the front end main frame can only be seen virtual disk VD, and can't see actual equipment, so being dealt into what comprise in the SCSI read write command on the VIC all is the logical space address of each virtual disk, and these addresses must be according to the PS list information of corresponding virtual disk VD in the AXUM metadata, be converted into the physical device physical address corresponding, and the order that comprises physical address is sent to corresponding device.Because the one section data that will read and write may be crossed over the multistage PS space (especially in the disk of striping tissue) of VD, therefore a logical order may be split into a plurality of physics orders.The target of address spaces work is to determine that each logical order will split into several actual command, and calculates the offset address and the data length of each actual command correspondence.Simultaneously, read-write operation also may need respectively the data of handling are assembled or split.
The STML module uses dummy command processing threads (VCPT) and physics command process thread (PCPT) to control state, disposal route and the flow process (Fig. 9) of each order fully respectively.The following describes the processing procedure of SCSI read write command in the STML module that the front end main frame is sent.
Order receiver function rx_cmnd () is after receiving a dummy command, generate a corresponding dummy command structure Target_Virtual_Scsi_Cmnd, and to put coomand mode be ST_NEW_CMND, then this structure inserted the dummy command formation, wakes the VCPT thread up.The VCPT thread is that the dummy command of ST_NEW_CMND calls dummy command analytic function analyse_virtual_cmd () to state, and this function is at the different operation of different command execution:
1) read command is handled:
For each physics order, fill a physics imperative structures Target_Physical_Scsi_Command.Is the state of this dummy command ST_DISASSEMBEL, and the state of putting corresponding each physics order is ST_PHYSICAL_NEW_CMND.
Physics command queue is inserted in these physics orders, and wake up process PCPT thread is handled.And the state of putting dummy command is ST_PROCESSING.(a)
PCPT thread dispatching physics command processing function handle_physical_cmd () handles actual command:
The state of physics order is changed to ST_PHYSICAL_CMD_PROCESSING.
After the scsi_do_req order is returned, call to return and handle function te_cmd_processed_new ().This function at first state of glove reason order is ST_PHYSICAL_CMD_DONE, judges whether that then all physics orders of the dummy command that this physics order is affiliated all are finished.If all physics orders of a dummy command all are finished, the state of then putting dummy command is ST_ASSEMBLE, and the thread VCPT of wake up process dummy command formation.
VCPT is the dummy command of ST_ASSEMBLE for state, calls data pooled function simulate (), collects the response of all physics orders of this dummy command correspondence, forms the response of dummy command then.For read operation, also to the data that each physics order is returned be merged.The state of rearmounted dummy command is ST_DONE.
After HBA drove call instruction end process function scsi_target_done (), VCPT changed over ST_DEQUEUE to the state of dummy command, and all physics coomand modes of putting this dummy command correspondence are ST_PHYSICAL_CMD_DEQUEUE.
The VCPT thread is that the dummy command of ST_DEQUEUE is carried out following operation for state: earlier this dummy command corresponding physical order is deleted from respective queue, then this dummy command is deleted from the dummy command formation.
2) write order is handled:
Physics command queue is inserted in the physics order.
The HBA driving is called Data Receiving function scsi_rx_data () and is given VCPT the data that transmit, and the state of dummy command is changed to ST_TO_PROCESS.And the state of each physics order is changed to ST_PHYSICAL_CMD_TO_PROCESS.
7. function is summed up
The functional module of AXUM system only realizes on the VIC node, can directly carry out on VIC the configuration effort of mass memory resource, also can be undertaken by the management software on the telemanagement main frame.AXUM adds virtualization system to each equipment in the storage networking one by one and manages, and utilizes the storage space of these equipment to create virtual disk.For existing virtual disk, can delete, capacity is revised and host assignment.The front end main frame finds to distribute to its virtual disk by the driver that loads optical channel card, and can as format, create disk file system etc. as using common scsi disk that virtual disk is carried out various operations.Main frame uses the SCSI standard commands to the visit of virtual disk, in the STML module on VIC, dummy command is shone upon, make them be converted into physics order towards actual physical device, thereby realize the correct conversion processing of logical space, guarantee that the operation and the function of virtual unit is correctly reliable to physical space.
Correlated performance test of the present invention
The virtual disk readwrite performance that VIC offers front end as shown in figure 10.Horizontal ordinate is the burst number (stripe number) of virtual disk VD among the figure, and ordinate unit is per second data throughout (MB/s).From test result as seen, along with the increase of stripe quantity, the performance of respective virtual dish also can improve gradually.Under equal conditions, front end host access virtual disk, even this VD does not carry out burst (stripe number=1), also will be than the I/O performance boost about 7% of physical storage devices in the network.Such performance gap mainly be since the internal memory of the VIC node high capacity (1GB) that can provide as Data Cache, thereby the VD readwrite performance that makes AXUM provide is better than generic storage equipment.Simultaneously, if use band organizational form (stripe), the performance advantage that the virtual disk that AXUM provides brings will be more obvious.
The I/O access delay of process virtualization process as shown in figure 11.Ordinate is represented the average response time (unit is ms) of I/O operation among the figure.From test result as can be seen, because the transmission and the FC protocol processes of optical-fibre channel can be brought delay, so the response time outline of the network equipment is longer than local disk (about 9%).But be the network equipment equally, the response time of the virtual disk that the AXUM system provides has only Duoed about 0.8% than the straight-forward network disk access, as seen, the software delays that is caused by the AXUM virtualization system (mainly comprising the dismounting of map addresses calculating and data) is very little.
The data-handling capacity of VIC node as shown in figure 12.The different virtual dish of our carry AXUM establishment on every front end main frame, each VD only takies the space of a PD, and feasible visit to the different virtual dish can be independent of each other.Use the Iometer testing tool that these virtual disks are carried out complete sequential read operational testing, so that obtain the VIC maximum data throughput of no buffer memory influence.From test result as can be seen, during less than 120MB/s, the processing power of VIC and the operating load of front end are the relation of linear increment substantially at data throughput.But after data throughput was greater than 120MB/s, the processing power of VIC was tending towards saturated, and the average I/O response time also significantly increases.This is because (the actual maximum data throughput of 160MB bandwidth SCSI bus is about 135MB) that the bandwidth constraints of SCSI bus causes, and the maximum data processing power of VIC node is about 127MB/s.And this moment, cpu busy percentage was less than 20%.As seen, under existing hardware condition, the processing power of VIC depends primarily on the performance of each parts on the data path, rather than the computing power of VIC node.In order to eliminate the performance bottleneck of VIC, the most effective means can only be the quantity of extension process node, make a plurality of node cooperative works, thereby the operating load that can adapt to storage networking changes.
Claims (1)
- Based on network large scale resource memory managing method is characterized in that in the SAN environment, and it is at following storage area network, promptly realizes among the SAN, and described network contains:Main frame, promptly front-end server has more than 1, and wherein one as management host, the installation administration program interface;Optical fiber switch, it and each main frame, node machine and optical fiber memory device interconnect;
The node machine, represent to have only one with VIC, being inserted with an optical fiber driver card starter pattern and target mode is each one of FC-HBA, it links to each other with described optical fiber switch, and resource distribution module, software object device module and management software Agent are installed;
The rear end memory device comprises that the SCSI that direct optical fiber disk that links to each other with described optical fiber switch and the common I/O node of process link to each other with optical fiber switch is a minicomputer interface system disk array;
In described disk, two metadata disks are arranged, be used to preserve all metadata informations, described metadata is meant in user's use, disc information after the each change that forms during the operation of the establishment that comprises physical disk, virtual deposit that takes place, deletion, expansion, described metadata comprises following content:The basic metadata of PD position comprises that the overall situation of stored physical equipment is unified identifier, represent with UUID, and type information is interior;The basic metadata of SC, SC is meant the storage and uniform space that various memory devices are formed, the basic metadata of SC is the data of SC message structure in the system;The metadata of all PD of SC, i.e. all actual memory devices among the SC, i.e. the UUID information of all PD;The metadata of the VD that SC is all comprises not only that promptly all VD are the data of virtual disk message structure, also comprise each VD in actual disk shared position, size in interior information; Except that the metadata dish, all the other disks of system are only preserved the basic metadata of PD separately;
When method of the present invention was implemented in above-mentioned storage area network, it contained following steps successively:The 1st step: at described I/O node machine is to load in the VIC node with lower module:The resource distribution module is represented with axum.o; Its definite virtual disk parameter that will create, comprise the title, size of virtual disk, whether band tissue and data fragmentation number, interior nuclear symbol, kernel information interface are provided simultaneously and register special character type equipment, so that described module can receive user's virtual configuration-direct, and revise the respective resources configuration information that it is safeguarded, simultaneously the information that changes is saved in the metadata dish;Scsi target device module is represented with mid_target.o, and it organizes the scsi device formation according to the configuration information of axum.o, sets up the command process thread simultaneously, and the I/O that sends in order to processing asks; For this reason, it sets up the scsi command that the thread of dummy command and physics order handles at virtual disk respectively is dummy command, and the scsi command for actual physical disks after dummy command decomposed is the chain phenotype formation of physics order;When the above-mentioned I/O of processing asks, carry out the fractionation or the merging processing of map addresses and data according to the configuration information of axum.o, return result;The 2nd step: resource distribution; The keeper carries out virtual configuration by the configuration tool that the node machine provides to the mass memory resource, and it contains following steps successively:The 2.1st step: the user is by the definite virtual disk parameter that will create of resources configuration management software;The 2.2nd step: the axum.o module finds that by ioctl mechanism the notice kernel that user's space transmits carries out the command id that virtual disk is created, the parameter that will pass to user's space kernel is taken out, comprise that virtual disk is title, size, organizational form and the strip data burst number of VD, carry out virtual disk and create;The 2.3rd step: creating the virtual disk structure is the variable of ax_vd_t, inserts parameter information, and writes down new virtual disk configuration data;The 2.4th step: judge the organizational form in the parameter, carry out corresponding allocation of space and handle:The 2.4.1 step: when being connective tissue's mode:Determine at first whether the virtual disk in the parameter is the new disk of creating, if not, then preferentially select the affiliated physical equipment of final stage physical space of this disk to carry out spatial spread; Otherwise, preferentially use best physical disk to carry out allocation of space;Secondly, determine the space size that residue will be distributed, select new disk to carry out allocation of space;The allocation of space success is returned;The 2.4.2 step: when being the band organizational form, suppose that band quantity is s:At first, determine whether to be new virtual disk, if then, select s physical disk to carry out allocation of space to all physical equipment orderings; Otherwise, select in original virtual disk the last pairing physical disk of s piece storage space to carry out allocation of space;The space size that definite residue will be distributed;Upgrade the capacity of virtual disk and the size of remaining space;To the physical equipment rearrangement, select s new equipment;The allocation of space success is returned;The 2.5th step: fill new virtual disk structure, and this structure is inserted in the global variable, notify user's virtual disk to create and finish;The 3rd step: metadata is preserved, and it contains following steps successively:When operating, at first write the basic metadata of PD in the basic metadata of operated physical disk PD position at physical disk;The position of preserving the basic metadata of SC in the metadata dish writes the information data of SC;The position of preserving all VD metadata of SC in the metadata dish writes the basic metadata of VD and the occupied information of VD;The 3rd step: data accessThe 3.1st step: mid_target.o receives the SCSI read write command to the virtual disk logical space that front-end server is sent through interface function, and generates the corresponding instruction structure, coomand mode is set after, these imperative structures are inserted the processing queue of dummy command;The 3.2nd step: mid_target.o wakes the dummy command thread up, calls Virtual Analysis instruction, judges and at the different operation of different command execution;The 3.3rd step: read command is handled, and comprises following steps successively:According to the virtual configuration information that the axum.o module is safeguarded, the mapping function of using by resource distribution is mapped to several physics scsi commands to dummy command;To each physics order, make up the corresponding instruction structure, simultaneously, the state of dummy command and the state of each physics order are set;The formation of physics command process is inserted in these physics orders, wake physics command process thread up and handle, and the respective handling state of dummy command is set;After the physics command process finishes, the dummy command processing threads is collected the response of all physics orders, forms the response of dummy command, and the data that each physics order is returned merge simultaneously, then the response of this dummy command is sent to the optical fiber driver card, and revise a coomand mode;The dummy command processing threads is deleted the corresponding physical order from respective queue, this dummy command is deleted from the dummy command processing queue again;The 3.4th step: write order is handled, and comprises following steps successively:According to the virtual configuration information that the axum.o module is safeguarded, the mapping function of using by resource distribution is mapped to several physics orders to dummy command;For each physics order, make up the corresponding instruction structure, simultaneously, the state of dummy command and the state of each physics order are set;The formation of physics command process is inserted in these physics orders;The dummy command processing threads drives from the front-end server forward data according to the state calling interface function notice sonet card of dummy command, resets the state of dummy command and corresponding each physics order;Sonet card drives and gives the dummy command processing queue data by interface function, and dummy command and each physics order are set as the state of handling;The dummy command processing threads data fragmentation and be dealt in the corresponding physics order, is reset the state of dummy command and each physics order by interface function;Physics command process thread is handled;After disposing, drive notice front end main frame by sonet card, and the state of each physics order of change dummy command and correspondence, the dummy command processing threads is deleted the corresponding physical order from respective queue, this dummy command is deleted from the dummy command processing queue again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100112313A CN1299207C (en) | 2005-01-21 | 2005-01-21 | Large scale resource memory managing method based on network under SAN environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100112313A CN1299207C (en) | 2005-01-21 | 2005-01-21 | Large scale resource memory managing method based on network under SAN environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1645342A true CN1645342A (en) | 2005-07-27 |
| CN1299207C CN1299207C (en) | 2007-02-07 |
Family
ID=34875498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100112313A Expired - Fee Related CN1299207C (en) | 2005-01-21 | 2005-01-21 | Large scale resource memory managing method based on network under SAN environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1299207C (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100353307C (en) * | 2006-02-16 | 2007-12-05 | 杭州华三通信技术有限公司 | Storage system and method of storaging data and method of reading data |
| WO2007143882A1 (en) * | 2006-06-09 | 2007-12-21 | Fast Dragon International Holdings Limited | N^n data management, access, storage, transfer, exchange and retrieval system (data master) |
| CN100428131C (en) * | 2006-10-13 | 2008-10-22 | 清华大学 | Method for distributing resource in large scale storage system |
| CN100451976C (en) * | 2007-07-23 | 2009-01-14 | 清华大学 | Migration management based on massive data classified memory system |
| CN100464535C (en) * | 2006-05-09 | 2009-02-25 | 国家数字交换系统工程技术研究中心 | Dynamic loading and control method for router virtual drive module |
| CN1904855B (en) * | 2005-07-29 | 2010-05-26 | 国际商业机器公司 | System and method for automatically relating components of a storage area network in a volume container |
| CN101510145B (en) * | 2009-03-27 | 2010-08-25 | 杭州华三通信技术有限公司 | Storage system management method and apparatus |
| CN101815095A (en) * | 2010-03-26 | 2010-08-25 | 山东高效能服务器和存储研究院 | SAN stored resource unified management and distribution method |
| US8190846B2 (en) | 2005-10-04 | 2012-05-29 | Hitachi, Ltd. | Data management method in storage pool and virtual volume in DKC |
| CN102868684A (en) * | 2012-08-28 | 2013-01-09 | 浪潮(北京)电子信息产业有限公司 | Fiber channel target and realizing method thereof |
| CN102917005A (en) * | 2012-08-28 | 2013-02-06 | 大唐软件技术股份有限公司 | Method and device supporting massive memory access to transactions |
| CN103701830A (en) * | 2014-01-13 | 2014-04-02 | 浪潮(北京)电子信息产业有限公司 | Method and system for realizing data processing and interaction |
| CN104461698A (en) * | 2014-12-29 | 2015-03-25 | 成都致云科技有限公司 | Dynamic virtual disk mounting method, virtual disk management device and distributed storage system |
| CN104536845A (en) * | 2014-12-10 | 2015-04-22 | 深圳市锐明视讯技术有限公司 | Write cache data synchronization method and device |
| CN104657149A (en) * | 2015-03-13 | 2015-05-27 | 浪潮集团有限公司 | Software framework implementation method of management module of storage system |
| CN105653529A (en) * | 2014-11-12 | 2016-06-08 | 富泰华工业（深圳）有限公司 | Storage management system, device and method |
| CN105786777A (en) * | 2016-02-22 | 2016-07-20 | 中国移动通信集团广东有限公司 | Storage capacity centralized processing method based on text parser |
| CN106155948A (en) * | 2015-04-16 | 2016-11-23 | 伊姆西公司 | For storing the method for system, aggregation configuration engine, aggregation configuration instrument and file system |
| CN103744790B (en) * | 2005-08-25 | 2017-05-10 | 美国莱迪思半导体公司 | Smart scalable storage switch architecture |
| CN108052393A (en) * | 2017-12-27 | 2018-05-18 | 广州烽睿信息科技有限公司 | A kind of IT resource pools intelligent optimization configuration device |
| CN111857598A (en) * | 2020-07-28 | 2020-10-30 | 山东超越数控电子股份有限公司 | Mapping method, device, equipment and readable medium for remote disk localization |
| CN113472623A (en) * | 2021-05-31 | 2021-10-01 | 山东英信计算机技术有限公司 | Storage system management method, device, storage medium and equipment |
| CN113535088A (en) * | 2021-07-15 | 2021-10-22 | 深圳市安存数据技术有限公司 | Storage device, data processing method and storage system |
| CN113760525A (en) * | 2020-06-03 | 2021-12-07 | 百度（美国）有限责任公司 | Data protection with static resource partitioning for data processing accelerators |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6715031B2 (en) * | 2001-12-28 | 2004-03-30 | Hewlett-Packard Development Company, L.P. | System and method for partitioning a storage area network associated data library |
| CN1254748C (en) * | 2003-10-31 | 2006-05-03 | 清华大学 | Method for accessing distributed and virtualized storage in local network |
| CN1258713C (en) * | 2003-11-14 | 2006-06-07 | 清华大学 | Data distribution dynamic mapping method based on magnetic disc characteristic |
-
2005
- 2005-01-21 CN CNB2005100112313A patent/CN1299207C/en not_active Expired - Fee Related
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1904855B (en) * | 2005-07-29 | 2010-05-26 | 国际商业机器公司 | System and method for automatically relating components of a storage area network in a volume container |
| CN103744790B (en) * | 2005-08-25 | 2017-05-10 | 美国莱迪思半导体公司 | Smart scalable storage switch architecture |
| US8683171B2 (en) | 2005-10-04 | 2014-03-25 | Hitachi, Ltd. | Data management method in storage pool and virtual volume in DKC |
| US8190846B2 (en) | 2005-10-04 | 2012-05-29 | Hitachi, Ltd. | Data management method in storage pool and virtual volume in DKC |
| US9448733B2 (en) | 2005-10-04 | 2016-09-20 | Hitachi, Ltd. | Data management method in storage pool and virtual volume in DKC |
| CN100353307C (en) * | 2006-02-16 | 2007-12-05 | 杭州华三通信技术有限公司 | Storage system and method of storaging data and method of reading data |
| CN100464535C (en) * | 2006-05-09 | 2009-02-25 | 国家数字交换系统工程技术研究中心 | Dynamic loading and control method for router virtual drive module |
| WO2007143882A1 (en) * | 2006-06-09 | 2007-12-21 | Fast Dragon International Holdings Limited | N^n data management, access, storage, transfer, exchange and retrieval system (data master) |
| CN100428131C (en) * | 2006-10-13 | 2008-10-22 | 清华大学 | Method for distributing resource in large scale storage system |
| CN100451976C (en) * | 2007-07-23 | 2009-01-14 | 清华大学 | Migration management based on massive data classified memory system |
| CN101510145B (en) * | 2009-03-27 | 2010-08-25 | 杭州华三通信技术有限公司 | Storage system management method and apparatus |
| CN101815095B (en) * | 2010-03-26 | 2016-01-13 | 浪潮电子信息产业股份有限公司 | A kind of SAN stored resource unified management and distribution method |
| CN101815095A (en) * | 2010-03-26 | 2010-08-25 | 山东高效能服务器和存储研究院 | SAN stored resource unified management and distribution method |
| CN102917005A (en) * | 2012-08-28 | 2013-02-06 | 大唐软件技术股份有限公司 | Method and device supporting massive memory access to transactions |
| CN102868684A (en) * | 2012-08-28 | 2013-01-09 | 浪潮(北京)电子信息产业有限公司 | Fiber channel target and realizing method thereof |
| CN102917005B (en) * | 2012-08-28 | 2016-10-26 | 大唐软件技术股份有限公司 | A kind of mass memory access method supporting affairs and device |
| CN103701830B (en) * | 2014-01-13 | 2016-09-07 | 浪潮(北京)电子信息产业有限公司 | A kind of data that realize process and mutual method and system |
| CN103701830A (en) * | 2014-01-13 | 2014-04-02 | 浪潮(北京)电子信息产业有限公司 | Method and system for realizing data processing and interaction |
| CN105653529A (en) * | 2014-11-12 | 2016-06-08 | 富泰华工业（深圳）有限公司 | Storage management system, device and method |
| CN104536845A (en) * | 2014-12-10 | 2015-04-22 | 深圳市锐明视讯技术有限公司 | Write cache data synchronization method and device |
| CN104536845B (en) * | 2014-12-10 | 2018-02-23 | 深圳市锐明技术股份有限公司 | The method of data synchronization and device of write buffer |
| CN104461698A (en) * | 2014-12-29 | 2015-03-25 | 成都致云科技有限公司 | Dynamic virtual disk mounting method, virtual disk management device and distributed storage system |
| CN104657149A (en) * | 2015-03-13 | 2015-05-27 | 浪潮集团有限公司 | Software framework implementation method of management module of storage system |
| CN106155948A (en) * | 2015-04-16 | 2016-11-23 | 伊姆西公司 | For storing the method for system, aggregation configuration engine, aggregation configuration instrument and file system |
| CN106155948B (en) * | 2015-04-16 | 2019-07-05 | 伊姆西公司 | For the method for storage system, aggregation configuration engine, aggregation configuration tool and file system |
| CN105786777A (en) * | 2016-02-22 | 2016-07-20 | 中国移动通信集团广东有限公司 | Storage capacity centralized processing method based on text parser |
| CN105786777B (en) * | 2016-02-22 | 2018-09-28 | 中国移动通信集团广东有限公司 | Memory capacity based on text resolver focuses on method |
| CN108052393A (en) * | 2017-12-27 | 2018-05-18 | 广州烽睿信息科技有限公司 | A kind of IT resource pools intelligent optimization configuration device |
| CN113760525A (en) * | 2020-06-03 | 2021-12-07 | 百度（美国）有限责任公司 | Data protection with static resource partitioning for data processing accelerators |
| CN111857598A (en) * | 2020-07-28 | 2020-10-30 | 山东超越数控电子股份有限公司 | Mapping method, device, equipment and readable medium for remote disk localization |
| CN113472623A (en) * | 2021-05-31 | 2021-10-01 | 山东英信计算机技术有限公司 | Storage system management method, device, storage medium and equipment |
| CN113535088A (en) * | 2021-07-15 | 2021-10-22 | 深圳市安存数据技术有限公司 | Storage device, data processing method and storage system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1299207C (en) | 2007-02-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1299207C (en) | Large scale resource memory managing method based on network under SAN environment | |
| CN100533387C (en) | System and method for executing job step | |
| JP5244236B2 (en) | Computer system, method, and program | |
| CN1220946C (en) | Method of partition block for managing computational environment, system and program product | |
| US7480719B2 (en) | Information system, load control method, load control program and recording medium | |
| CN1881183A (en) | Information processing device, procedure control method and computer program | |
| CN1722107A (en) | Computer system and method for migrating from one storage system to another | |
| CN1246778C (en) | Information processing system and distribution method of storage device | |
| CN1928840A (en) | Control method of device in storage system for virtualization | |
| CN1581877A (en) | Storage system | |
| CN1300723C (en) | System and method for virtual devices using a plurality of processors | |
| CN1453715A (en) | Storing system having vitual source | |
| CN1265274C (en) | Storage device control device and control method of storage device control device | |
| CN1791862A (en) | Operating systems | |
| CN1836212A (en) | Automatic and dynamic provisioning of databases | |
| CN1955940A (en) | RAID system, RAID controller and rebuilt/copy back processing method thereof | |
| US20100125715A1 (en) | Storage System and Operation Method Thereof | |
| CN1608257A (en) | Aggregate system resource analysis including correlation matrix and metric-based analysis | |
| CN101042676A (en) | Storage system, storage extent release method and storage apparatus | |
| CN1324472C (en) | Information providing system | |
| JP6172649B2 (en) | Information processing apparatus, program, and information processing method | |
| JP5385912B2 (en) | Calculation device, system management device, calculation method, and program | |
| JP4801761B2 (en) | Database management method and system, and processing program therefor | |
| CN1825267A (en) | Storage controller and controlling method therefor | |
| CN1838128A (en) | Control technology for storage system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20070207 Termination date: 20110121 |