CN108268210A - A kind of information processing method, calculate node and memory node - Google Patents

A kind of information processing method, calculate node and memory node Download PDF

Info

Publication number
CN108268210A
CN108268210A CN201611270149.7A CN201611270149A CN108268210A CN 108268210 A CN108268210 A CN 108268210A CN 201611270149 A CN201611270149 A CN 201611270149A CN 108268210 A CN108268210 A CN 108268210A
Authority
CN
China
Prior art keywords
node
memory node
request
data
volume
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
Application number
CN201611270149.7A
Other languages
Chinese (zh)
Other versions
CN108268210B (en
Inventor
陈方县
沈玉良
王东旭
孟令坤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Mobile Communications Group Co Ltd
China Mobile Suzhou Software Technology Co Ltd
Original Assignee
China Mobile Communications Group Co Ltd
China Mobile Suzhou Software Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Mobile Communications Group Co Ltd, China Mobile Suzhou Software Technology Co Ltd filed Critical China Mobile Communications Group Co Ltd
Priority to CN201611270149.7A priority Critical patent/CN108268210B/en
Publication of CN108268210A publication Critical patent/CN108268210A/en
Application granted granted Critical
Publication of CN108268210B publication Critical patent/CN108268210B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
    • G06F3/0604—Improving or facilitating administration, e.g. storage management
    • G06F3/0607—Improving or facilitating administration, e.g. storage management by facilitating the process of upgrading existing storage systems, e.g. for improving compatibility between host and storage device
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
    • G06F3/0614—Improving the reliability of storage systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0655—Vertical data movement, i.e. input-output transfer; data movement between one or more hosts and one or more storage devices
    • G06F3/0656—Data buffering arrangements
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
    • G06F3/067—Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The embodiment of the invention discloses a kind of information processing method, this method includes：When using the first data link processes read-write requests, detect the connection status of the first data link, the link that first data link is established for the first memory node with calculate node, the first memory node are memory node currently in use corresponding with the first volume that calculate node uses in memory node cluster；If detecting, the first data link is in off-state, it then sends and updates the data request to the second memory node, request is updated the data for asking the current meta data information of the download first volume, second memory node is the slack storage node corresponding with the first volume that calculate node uses in memory node cluster in addition to the first memory node；It receives the response of the second memory node and updates the data the update response message that request is sent；If update response message is is updated successfully, transmission handovers request to the second memory node.The embodiment of the present invention further simultaneously discloses a kind of calculate node and memory node.

Description

A kind of information processing method, calculate node and memory node
Technical field
The present invention relates to the memory technology in computer realm more particularly to a kind of information processing method, calculate node and Memory node.
Background technology
It is highlighted with universal and big data, the development of cloud computing, the tired state of conventional memory array of virtualization, it is distributed Processing capacity, autgmentability, versatility, reliability etc. are stored in respect to for conventional memory array, there is more obvious advantage.
At present, in distributed memory system, to ensure business continuance, high-availability cluster is provided to user, according to more Path technique causes any one volume in distributed memory system all by least being controlled there are two memory node.Specifically , it will be rolled up by each memory node and calculate node exported to by network, so as to construct a data link respectively.Calculate node It can be directed to per data one virtual volume of link establishment, so as to be created that at least two virtual volumes, and this at least two virtual volume Same volume in corresponding distributed memory system, in this case, calculate node can utilize multipath technology to this at least two A virtual volume is managed, and then calculate node and each memory node of this any one volume is controlled to constitute a data Link.In active-standby mode, calculate node can select wherein a data link processing I/O request by multipath technology, It is main memory node that usually the memory node for handling I/O request, which is known as, and other memory nodes are standby memory node, when Primary storage node failure, multipath detect the data link being connect with primary storage node and interrupt, I/O request will be cut in time Standby memory node is changed to, so as to ensure not interrupting for customer service.
However, the complexity of the distributed storage due to memory node, the control in some distributed system memory nodes Subsystem needs to record the metadata information of volume, therefore, by its metadata information active load to control when memory node export is rolled up In subsystem.In this way when the volume has I/O request, the multipath of calculate node can select one of memory node as master Memory node, the corresponding control subsystem of primary storage node handle I/O request according to the volume metadata information for most starting load, and And the metadata information of volume can be constantly updated during I/O request is handled.When primary storage node failure, calculate node is led to Cross multipath and I/O request be switched to standby memory node, the metadata information of volume that the control subsystem of standby memory node records this When do not updated actively, still with the metadata information of the volume of most start recording go processing I/O request.In this case, If the metadata information of volume has changed, this necessarily will appear serious data correctness problem, at I/O request Manage mistake.
Invention content
In order to solve the above technical problems, it an embodiment of the present invention is intended to provide a kind of information processing method, calculate node and deposits Node is stored up, can ensure the normal processing of read-write requests, improve the accuracy rate of read-write requests.
The technical proposal of the invention is realized in this way：
An embodiment of the present invention provides a kind of information processing method, applied in calculate node, including：
When using the first data link processes read-write requests, the connection status of first data link is detected, it is described The link that first data link is established for the first memory node with the calculate node, first memory node are memory node Memory node currently in use corresponding with the first volume that the calculate node uses in cluster；
If detecting, first data link is in off-state, and transmission, which updates the data, asks to the second memory node, Described to update the data request for asking the current meta data information of the download first volume, second memory node is described It is corresponding with the first volume that the calculate node uses spare to deposit in memory node cluster in addition to first memory node Store up node；
It receives and updates the data the update response message that request is sent described in the second memory node response；
If for the update response message to be updated successfully, transmission handovers request to second memory node, described to cut Request is changed to be used to the link for handling read-write requests being switched to the second data link, second number from first data link The link established according to link for second memory node with the calculate node.
Optionally, described receive updates the data the update response message that request is sent described in the second memory node response Later, the method further includes：
If the update response message fails for update, transmission is forbidden to handover request to second memory node.
Optionally, after the connection status of detection first data link, the method further includes：
If detect that first data link is in connected state, it is continuing with first data link processes and reads Write request.
Optionally, described using before the first data link processes read-write requests, the method further includes：
Receive the first read write command；
First read write command is responded, storage resource is sent and asks to the memory node cluster；
It receives the memory node cluster and responds the storage resource response message that the storage resource request is sent, it is described to deposit Corresponding first memory node of the first volume, the first volume and described second is carried in storage resource response message to deposit Store up node.
The embodiment of the present invention additionally provides a kind of information processing method, applied in the second memory node, including：
Receive calculate node transmission updates the data request；
Request is updated the data described in response, the current meta data information of the first volume is downloaded in request, and the first volume is described The storage unit that calculate node uses；
If the current meta data information of the first volume, which is downloaded, successfully, sends the update response message that is updated successfully to institute State calculate node；
The handover request that the calculate node is sent is received, is read according to the handover request using the second data link processes Write request, the link that second data link is established for second memory node with the calculate node.
Optionally, request is updated the data described in the response, after the current meta data information of the first volume is downloaded in request, institute The method of stating further includes：
If the current meta data information failed download of the first volume, the update response message of update failure is sent to institute State calculate node.
An embodiment of the present invention provides a kind of calculate node, including：
Detection unit, for when using the first data link processes read-write requests, detecting first data link Connection status, the link that first data link is established for the first memory node with the calculate node, first storage Node is memory node currently in use corresponding with the first volume that the calculate node uses in memory node cluster；
First transmitting element, if for detecting that first data link is in off-state, transmission updates the data please It asks to the second memory node, the current meta data information for updating the data request and the first volume being downloaded for request, it is described Second memory node be in the memory node cluster in addition to first memory node, use with the calculate node described in The corresponding slack storage node of the first volume；
First receiving unit, for receiving the update sound that request is updated the data described in the second memory node response and is sent Answer message；
First transmitting element, if being additionally operable to the update response message to be updated successfully, transmission handovers request to Second memory node, the handover request are used to the link for handling read-write requests being switched to from first data link Second data link, the link that second data link is established for second memory node with the calculate node.
Optionally, the calculate node further includes：Forbid unit；
It is described to forbid unit, for described receive request transmission is updated the data described in the second memory node response more Newly after response message, if the update response message fails for update, transmission is forbidden to handover request to second storage Node.
Optionally, the calculate node further includes：Read-write cell；
The read-write cell, after the connection status for detecting first data link, if detecting described When first data link is in connected state, then the first data link processes read-write requests are continuing with.
Optionally, first receiving unit is additionally operable to before the first data link processes read-write requests of the use, connects Receive the first read write command；
First transmitting element is additionally operable to respond first read write command, sends storage resource and asks to described to deposit Store up node cluster；
First receiving unit is additionally operable to receive what the memory node cluster response storage resource request was sent It is corresponding described to carry the first volume, the first volume in the storage resource response message for storage resource response message First memory node and second memory node.
An embodiment of the present invention provides a kind of second memory node, including：
Second receiving unit, for receive calculate node transmission update the data request；
Download unit, it is described for responding the current meta data information for updating the data request, asking to download the first volume The first volume is the storage unit that the calculate node uses；
Second transmitting element, if the current meta data information for the first volume is downloaded successfully, transmission is updated successfully Update response message to the calculate node；
Second receiving unit is additionally operable to receive the handover request that the calculate node is sent, please according to the switching It asks using the second data link processes read-write requests, second data link is saved for second memory node and described calculate The link that point is established.
Optionally, second transmitting element is additionally operable to update the data request described in the response, and the first volume is downloaded in request Current meta data information after, if the current meta data information failed download of the first volume, send update failure more New response message is to the calculate node.
An embodiment of the present invention provides a kind of information processing method, calculate node and memory node, when using the first data During link processing read-write requests, the connection status of the first data link is detected, the first data link is the first memory node and meter Operator node establish link, the first memory node be memory node cluster in it is corresponding with the first volume that calculate node uses The memory node used；If detecting, the first data link is in off-state, and transmission updates the data request to the second storage section Point updates the data request for asking the current meta data information of the download first volume, and the second memory node is memory node cluster In in addition to the first memory node, slack storage node corresponding with the first volume that calculate node uses；Receive the second memory node Response updates the data the update response message that request is sent；If update response message is is updated successfully, transmission handovers request to Second memory node, handover request are used to the link for handling read-write requests being switched to the second data-link from the first data link Road, the link that the second data link is established for the second memory node with calculate node.Using above-mentioned technic relization scheme, due to meter When detecting that a data link currently in use disconnects i.e. failure, which can first inform operator node The second memory node that will be switched to first carries out the update of the metadata information of the first volume, therefore, in the second memory node more After the new metadata information of the first volume, calculate node can use the in the corresponding data link of the second memory node A roll is written and read the processing of request, ensures that the normal processing of read-write requests in this way, and improves the accurate of read-write requests Rate.
Description of the drawings
Fig. 1 is a kind of system architecture diagram one of information processing system provided in an embodiment of the present invention；
Fig. 2 is a kind of flow chart one of information processing method provided in an embodiment of the present invention；
Fig. 3 is a kind of system architecture diagram two of information processing system provided in an embodiment of the present invention；
Fig. 4 is a kind of flowchart 2 of information processing method provided in an embodiment of the present invention；
Fig. 5 is a kind of flow chart 3 of information processing method provided in an embodiment of the present invention；
Fig. 6 is a kind of flow chart four of information processing method provided in an embodiment of the present invention；
Fig. 7 is the flow chart one of a kind of information processing method that the embodiment of the present invention also provides；
Fig. 8 is the flowchart 2 of a kind of information processing method that the embodiment of the present invention also provides；
Fig. 9 is a kind of interaction figure one of information processing method provided in an embodiment of the present invention；
Figure 10 is a kind of interaction figure two of information processing method provided in an embodiment of the present invention；
Figure 11 is a kind of interaction figure three of information processing method provided in an embodiment of the present invention；
Figure 12 is a kind of interaction figure four of information processing method provided in an embodiment of the present invention；
Figure 13 is a kind of structure diagram one of calculate node provided in an embodiment of the present invention；
Figure 14 is a kind of structure diagram two of calculate node provided in an embodiment of the present invention；
Figure 15 is a kind of structure diagram three of calculate node provided in an embodiment of the present invention；
Figure 16 is a kind of structure diagram four of calculate node provided in an embodiment of the present invention；
Figure 17 is a kind of structure diagram one of second memory node provided in an embodiment of the present invention；
Figure 18 is a kind of structure diagram two of second memory node provided in an embodiment of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete Site preparation describes.
A kind of information processing method provided in an embodiment of the present invention is realized based on the framework of an information processing system , as shown in Figure 1, information processing system can include 1 and at least two memory node of calculate node, 2 (i.e. memory node collection Group), wherein, front end forms calculate node by server (such as server without storage section), and rear end is by standard server (such as block storage server) forms at least two memory nodes 2.In above- mentioned information processing system, with a calculate node 1 For the application flow of a volume in memory node cluster 2, multiple memory nodes are managed using distributed system, structure Into distributed storage service cluster, and corresponding control subsystem is fitted in each memory node 2, in calculate node 1 Multipath subsystem is installed.Wherein, the multipath subsystem in calculate node 1 come control using any data link with storage Node 2 carries out information exchange by network.Specifically, calculate node will be stored by the distributed system in memory node cluster Then resource composition resource pool utilizes data link reality by establishing the first volume corresponding with calculate node in resource pool Existing information exchange between memory node and calculate node.Illustratively, calculate node establishes the first number with the first memory node According to link, the second data link is established with the second memory node, wherein, wherein, memory node cluster 2 has memory node 1, deposits Store up node 2 ..., memory node n compositions, illustratively with memory node 1 for the first memory node in Fig. 1, memory node 1 is Second memory node.
Embodiment one
An embodiment of the present invention provides a kind of information processing method, applied in calculate node, as shown in Fig. 2, this method It can include：
S101, when using the first data link processes read-write requests, detect the connection status of first data link, should The link that first data link is established for the first memory node with calculate node, first memory node are in memory node cluster Memory node currently in use corresponding with the first volume that calculate node uses.
Information processing method provided in an embodiment of the present invention is realized based on system architecture shown in FIG. 1, therefore, is calculated Node can establish out a plurality of data link into row information by the memory node in multipath system and memory node cluster Interaction.
Here, the first data link that calculate node can be established with the first memory node in storage cluster, and with Second memory node establishes corresponding second data link, wherein, the second memory node can be multiple, and therefore, second counts It can be corresponded to according to link multiple.Calculate node is by which data link of multipath system control selections into the friendship of row information Mutually.
In embodiments of the present invention, the first memory node can memory node, the second memory node can be standby deposit based on Node is stored up, the embodiment of the present invention is not restricted.Calculate node generally can first select primary storage node to correspond to by multipath system Data link into the interaction of row information, when primary storage node breaks down or data link is obstructed, can just be switched to standby deposit Store up interaction of the corresponding data link of node into row information.
It should be noted that in embodiments of the present invention, calculate node due to there is no storage section, the calculating section Point needs the volume in the resource pool by the corresponding memory node of data link to realize and reads and write when receiving read-write requests. That is, as shown in Figure 1, there are at least two memory nodes to correspond to the volume workable for a calculate node in resource pool, So as to form at least two data links.
Here, for calculate node using during the first data link processes read-write requests, which can be to this The connection status of first data link is monitored in real time, with judge the first data link currently in use with the presence or absence of failure or It goes wrong and is disconnected, wherein, the link which establishes for the first memory node and calculate node, this first is deposited Node is stored up as memory node currently in use corresponding with the first volume that calculate node uses in memory node cluster.
Optionally, in the embodiment of the present invention, data link can be FC (optical-fibre channel) data link, SAS data link Or iSCSI data link, the type of data link can be determined by the mode or technology of realization information exchange, the present invention is implemented Example is not restricted.
If S102, detecting that the first data link is in off-state, transmission, which updates the data, asks to the second memory node, This updates the data request for the current meta data information for asking the download first volume, which is memory node cluster In in addition to the first memory node, slack storage node corresponding with the first volume that calculate node uses.
After calculate node has detected the connection status of the first data link, which can know the first data-link The connection on road, the connection can there are two types of, one kind be in connection status, that is, connected state；One kind is in disconnection State, that is, off-state.If the first data link is in off-state, characterizes first data link and may have occurred failure, no It can be into the transmission or interaction of row information.In this case, calculate node just needs to realize the first number by other data link According to the information interactive process of link.
Here, if calculate node detects that the first data link is in off-state, which can send more New data is asked to the second memory node, this updates the data request for asking to change current member of second memory node to the first volume Data information is downloaded or updates, wherein, second memory node be memory node cluster in addition to the first memory node, with The corresponding slack storage node of the first volume that calculate node uses.
It should be noted that due to the slack storage node in the storage cluster in the embodiment of the present invention can be it is multiple, Therefore, the second memory node that realization here updates the data request can be any one in multiple slack storage nodes.
It is understood that since the first memory node and the second memory node are first workable for corresponding calculate node Volume, in this way, first memory node and the second memory node can know the corresponding initial element number of the first volume at the beginning It is believed that breath, wherein, metadata information is the storage information for characterizing the first volume, for example, store size, storage address and Store the information of service condition etc..When calculate node is written and read request using the first data link by the first volume, First memory node can update variation (such as the storage of the metadata information of the first volume in real time during read-write requests Location is where, stores the information such as how many space), thus can receive read or write data when, it is known that will These read or write the position that data write on the corresponding memory of the first volume.But calculate node using the first data link into During row read-write requests, the second memory node can or can not carry out the newer of the metadata information of the first volume, therefore, in the first data There is failure in link, when calculate node wants to switch to the second data link and is written and read the realization of request, i.e. calculate node When detecting that the first data link is in off-state, which, which can first send, updates the data request to the second storage section Point allows the second memory node first to go update to the metadata information of the newest first volume.In this way, subsequently using the second data-link When road is written and read request, the accuracy rate of read-write can be just improved.
S103, the update response message that the response of the second memory node updates the data request transmission is received.
After calculate node transmission updates the data request to the second memory node, which can be according to more New data request carries out having updated for the metadata information of the first volume, after second memory node update, can return to update and ring Answering message, the update response message characterizes the result that the second memory node updates the data to calculate node.
Optionally, in the embodiment of the present invention, the update result updated in response message can include：It is updated successfully or updates Failure.
Further, as shown in figure 3, calculate node can be provided with the sending module of notify subsystems, storage cluster In each memory node can set the receiving modules of notify subsystems.First in calculate node and storage cluster deposits It stores up node and the second memory node can be updated before calculate node sends handover request by notify subsystems The interaction of data-message and update response message.Wherein, memory node cluster have memory node 1, memory node 2 ..., deposit Node n compositions are stored up, Fig. 3 is illustrative, and with memory node 1 for the first memory node, memory node 1 is the second memory node.
If S104, update response message is are updated successfully, transmission handovers request to the second memory node, the handover request For the link for handling read-write requests to be switched to the second data link from the first data link, which is second The link that memory node is established with calculate node.
Calculate node receive the second memory node response update the data request send update response message after, due to Update response message in update result can there are two types of：It is updated successfully or updates failure.Therefore, if update response message is more New success, the second memory node of characterization have been updated to the corresponding newest metadata information of the first volume, and calculate node can be with It is written and read and requested using the second memory node, then, calculate node, which is then sent, handovers request to the second memory node, this is cut It changes request to be used to the link for handling read-write requests being switched to the second data link from the first data link, second data link The link established for the second memory node and calculate node.That is, calculate node can be started using the second data link It is written and read and requests.
It should be noted that the sending module of notify subsystems can be provided in calculate node, in storage cluster In the case that each memory node can set the receiving module of notify subsystems, multipath can be passed through in calculate node I/O request (read-write requests) can be forwarded in memory node rear end distributed volume (i.e. the first volume) by system.Calculate node is by multichannel Diameter is configured to active/passive patterns, ensures that any moment only has a data link (the first data link) in this way It is working, if the data link failure, I/O request will be switched to another by calculate node by multipath system Data link (the second data link), here crucially by the path of any I/O request of multipath Systematic selection it Before, either normal condition or other data link fault scenes, it is necessary to first pass through notify subsystems, carry out Handover request is first sent to notify-switch-send modules (hair by notify flows, calculate node by multipath system Send module), notify-switch-send forwards a request to notify-switch-receive (receiving module), notify- Under switch-receive is responsible for notifying the control subsystems (control subsystem of memory node) of the second memory node again Carry the metadata information of the first volume.For second memory node after update metadata information success, calculate node passes through multipath system System before new path switching occurs, can be not required to hold again using the path (the second data link) as I/O request path Row notify flows.
It is understood that the embodiment of the present invention need to only add in notify subsystems in original system architecture, so that it may To notify control subsystem load volumes of metadata information of active when multipath switches, and notify Subsystem logicals are simple, Its sending module is responsible for receiving multipath handover request, and receiving module is responsible for notifying control subsystem, is not related to too complicated patrol Volume, therefore, reduce the complexity of system design.Meanwhile do not need to add in management or arbitration node, the embodiment of the present invention carry The reliability of system entirety has been risen, has reduced the risk of system.
Further, as shown in figure 4, after S103, a kind of information processing method provided in an embodiment of the present invention further includes： S105.It is as follows：
If S105, update response message fail for update, transmission is forbidden to handover request to the second memory node.
Calculate node receive the second memory node response update the data request send update response message after, due to Update response message in update result can there are two types of：It is updated successfully or updates failure.Therefore, if update response message is more New failure, the second memory node of characterization do not update the corresponding newest metadata information of the first volume, and therefore, calculate node can be with Being written and read request using the second memory node can malfunction, and then, calculate node, which forbids sending, handovers request to the second storage section Point avoids the read-write error occurred using the second data link.
Further, if slack storage node has multiple, another can be carried out according to identical flow and spare is deposited Store up the data update and switching flow of node.
Further, as shown in figure 5, after S101, a kind of information processing method provided in an embodiment of the present invention further includes： S106.It is as follows：
If S106, when detecting that the first data link is in connected state, it is continuing with first data link processes reading Write request.
After calculate node has detected the connection status of the first data link, which can know the first data-link The connection on road, the connection can there are two types of, one kind be in connection status, that is, connected state；One kind is in disconnection State, that is, off-state.If the first data link is in connected state, characterizes first data link and failure does not have occurred, it can To continue the transmission of information or interaction.In this case, there is no need to realize by other data link for calculate node The information interactive process of one data link, therefore, calculate node can be continuing with the first data link processes read-write requests.
Further, as shown in fig. 6, before S101, a kind of information processing method provided in an embodiment of the present invention further includes： S107-S109.It is as follows：
S107, the first read write command is received.
S108, the first read write command of response, send storage resource and ask to memory node cluster.
S109, the storage resource response message that the response storage resource request of memory node cluster is sent, storage money are received Corresponding first memory node of the first volume, the first volume and the second memory node are carried in the response message of source.
In memory node cluster, the storage resource of a memory node is carried out by integration by distributed system and is collectively formed One storage pool, can be by creating the corresponding different calculate node of different volumes, so that calculating section in the storage pool Point can realize the write functionality of data by its corresponding volume.Specifically, it is asked when user is written and read by calculate node When asking, i.e., calculate node receives the first read write command, which calculates section to memory node cluster application storage resource Point the first read write command of response, sends storage resource and asks to memory node cluster.Then, memory node cluster is from storage pool Volume, that is, first volume of a specified size is created, the first volume is by globally shared, all memory nodes in memory node cluster It can access.The first volume is registered to control subsystems (control of the first memory node and the second memory node respectively System), the metadata information of the control subsystems record first volume, i.e. calculate node receive memory node cluster and ring during registration The storage resource response message of storage resource request transmission is answered, the first volume, the first volume are carried in the storage resource response message Corresponding first memory node and the second memory node.
Here, when calculate node is written and read request by multichannel strength system using the first data-link, in calculate node point The first volume is not imported from the control subsystems of two memory nodes (the first memory node and the second memory node), and respectively Two virtual volumes are generated in calculate node, the two virtual volumes copy each other is same in opposite rear end memory node cluster A volume (first volume), by the metamessage data register of the first volume to notify-switch-send during the first volume is imported With notify-switch-receive modules, calculate node is managed the two virtual volumes by multipath system, application Directly access multipath system.
An embodiment of the present invention provides a kind of information processing method, when using the first data link processes read-write requests, The connection status of the first data link is detected, the first data link is the link that the first memory node is established with calculate node, the One memory node is memory node currently in use corresponding with the first volume that calculate node uses in memory node cluster；If inspection It measures the first data link and is in off-state, then send and update the data request to the second memory node, update the data request and be used for Request download the first volume current meta data information, the second memory node be memory node cluster in addition to the first memory node, Slack storage node corresponding with the first volume that calculate node uses；It receives the response of the second memory node and updates the data request transmission Update response message；If update response message is is updated successfully, transmission handovers request to the second memory node, handover request For the link for handling read-write requests to be switched to the second data link from the first data link, the second data link is deposited for second Store up the link that node is established with calculate node.Using above-mentioned technic relization scheme, since calculate node is used detecting A data link disconnect i.e. break down when, the calculate node can first inform will be switched to second storage section Point first carries out the update of the metadata information of the first volume, therefore, the metadata information of the first volume is had updated in the second memory node Afterwards, calculate node can be written and read the processing of request using the first volume in the corresponding data link of the second memory node , the normal processing of read-write requests is ensured that in this way, and improves the accuracy rate of read-write requests.
Embodiment two
The embodiment of the present invention additionally provides a kind of information processing method, applied in the second memory node, as shown in fig. 7, This method can include：
What S201, reception calculate node were sent updates the data request.
Calculate node has detected the connection status of the first data link, which can know the first data link Connection, the connection can there are two types of, one kind be in connection status, that is, connected state；One kind is is off That is off-state.If the first data link is in off-state, characterizes first data link and may have occurred failure, it is impossible into The transmission or interaction of row information.In this case, calculate node just needs to realize the first data-link by other data link The information interactive process on road.Then, calculate node starts to find out the calculate node from memory node cluster corresponding spare Memory node, it is desirable to switch in read-write requests to slack storage node, the second memory node in the embodiment of the present invention is exactly standby With memory node, in this way, second memory node be received by calculate node transmission update the data request.
S202, response update the data request, and the current meta data information of the first volume is downloaded in request, which saves to calculate The storage unit that point uses.
The second memory node have received calculate node update the data request after, which can ring Request should be updated the data, request is gone to download the corresponding current newest first number of (acquisition) first volume from the first volume of resource pool It is believed that breath is current meta data information, wherein the first volume is the storage unit that calculate node uses.
It is understood that in embodiments of the present invention, since the first memory node and the second memory node are to accrued The first volume workable for operator node, in this way, first memory node and the second memory node can know at the beginning A roll of corresponding initial metadata information, wherein, metadata information is the storage information for characterizing the first volume, for example, depositing The information of size, storage address and the storage service condition of storage etc..Pass through the first volume using the first data link in calculate node When being written and read request, the first memory node can update the metadata letter of the first volume in real time during read-write requests The variation (such as storage address where, store the information such as how many space) of breath, can thus read or write receiving When data, it is known that these are read or write the position that data write on the corresponding memory of the first volume.But it is saved calculating When o'clock being written and read request using the first data link, the second memory node can or can not carry out the metadata information of the first volume more New, therefore, there is failure in the first data link, calculate node wants to switch to the second data link and is written and read request Realization when, i.e. for calculate node when detecting that the first data link is in off-state, which can first send update Request of data allows the second memory node first to go update to the metadata information of the newest first volume to the second memory node.In this way, When being subsequently written and read request using the second data link, the accuracy rate of read-write can be just improved.
If the current meta data information of S203, the first volume downloads the update response message for successfully, sending and being updated successfully extremely Calculate node.
The response of second memory node updates the data request, after the current meta data information of the first volume is downloaded in request, due to Current meta data information is downloaded in second memory node request to succeed, it is also possible to failure, therefore, second memory node life Into update response message in update result can be updated successfully or update failure, in this case, second storage Node is downloaded successfully in the current meta data information of the first volume, with regard to sending the update response message being updated successfully to calculate node.
Further, as shown in figure 3, calculate node can be provided with the sending module (notify- of notify subsystems Switch-send), each memory node in storage cluster can set the receiving module (notify- of notify subsystems switch-receive).The first memory node and the second memory node in calculate node and storage cluster can calculate Before node sends handover request, data-message is updated by notify subsystems and updates the interaction of response message. Wherein, memory node cluster by memory node 1, memory node 2 ..., memory node n form, Fig. 3 is illustrative, with storage Node 1 is the first memory node, and memory node 1 is the second memory node.
S204, the handover request that calculate node is sent is received, is read according to the handover request using the second data link processes Write request, the link which establishes for the second memory node with calculate node.
Second memory node is after the update response message to calculate node being updated successfully is sent, due to the second storage section Point has been downloaded or has updated the metamessage data for having arrived the newest first volume, and in this case, calculate node can pass through multichannel Diameter system carries out having switched for data link, and then, which receives the handover request that calculate node is sent, and root According to the handover request using the second data link processes read-write requests, which saves for the second memory node with calculating The link that point is established.
It should be noted that the sending module of notify subsystems can be provided in calculate node, in storage cluster In the case that each memory node can set the receiving module of notify subsystems, multipath can be passed through in calculate node I/O request (read-write requests) can be forwarded in memory node rear end distributed volume (i.e. the first volume) by system.Calculate node is by multichannel Diameter is configured to active/passive patterns, ensures that any moment only has a data link (the first data link) in this way It is working, if the data link failure, I/O request will be switched to another by calculate node by multipath system Data link (the second data link), here crucially by the path of any I/O request of multipath Systematic selection it Before, either normal condition or other data link fault scenes, it is necessary to first pass through notify subsystems, carry out Handover request is first sent to notify-switch-send modules (hair by notify flows, calculate node by multipath system Send module), notify-switch-send forwards a request to notify-switch-receive (receiving module), notify- Under switch-receive is responsible for notifying the control subsystems (control subsystem of memory node) of the second memory node again Carry the metadata information of the first volume.For second memory node after update metadata information success, calculate node passes through multipath system System before new path switching occurs, can be not required to hold again using the path (the second data link) as I/O request path Row notify flows.
It is understood that the embodiment of the present invention need to only add in notify subsystems in original system architecture, so that it may To notify control subsystem load volumes of metadata information of active when multipath switches, and notify Subsystem logicals are simple, Its sending module is responsible for receiving multipath handover request, and receiving module is responsible for notifying control subsystem, is not related to too complicated patrol Volume, therefore, reduce the complexity of system design.Meanwhile do not need to add in management or arbitration node, the embodiment of the present invention carry The reliability of system entirety has been risen, has reduced the risk of system.
Further, as shown in figure 8, after S202, a kind of information processing method that the embodiment of the present invention also provides also wraps It includes：S205.It is as follows：
If the current meta data information failed download of S205, the first volume, the update response message of update failure is sent extremely Calculate node.
The response of second memory node updates the data request, after the current meta data information of the first volume is downloaded in request, due to Current meta data information is downloaded in second memory node request to succeed, it is also possible to failure, therefore, second memory node life Into update response message in update result can be updated successfully or update failure, in this case, second storage Node is in the current meta data information failed download of the first volume, the update response message that just transmission update fails to calculate node .
The embodiment of the present invention additionally provides a kind of information processing method, and receive calculate node transmission updates the data request； Response updates the data request, and the current meta data information of the first volume is downloaded in request, and the first volume is deposited for what the calculate node used Storage unit；If the current meta data information of the first volume, which is downloaded, successfully, sends the update response message that is updated successfully to calculating section Point；The handover request that calculate node is sent is received, the second data link processes read-write requests, the second number are used according to handover request The link established according to link for second memory node with calculate node.Using above-mentioned technic relization scheme, saved due to calculating For point when detecting that a data link currently in use disconnects i.e. failure, which can first inform will The second memory node being switched to first carries out the update of the metadata information of the first volume, therefore, is had updated in the second memory node After the metadata information of the first volume, calculate node can use the first volume in the corresponding data link of the second memory node The processing of request is written and read, ensures that the normal processing of read-write requests in this way, and improves the accuracy rate of read-write requests.
Embodiment three
An embodiment of the present invention provides a kind of information processing method, as shown in figure 9, this method can include：
S301, calculate node receive the first read write command.
S302, calculate node respond the first read write command, send storage resource and ask to memory node cluster.
S303, calculate node receive the storage resource response message that the response storage resource request of memory node cluster is sent, Corresponding first memory node of the first volume, the first volume and the second memory node are carried in the storage resource response message.
The description during S301-S303 in the embodiment of the present invention and the description one of the S107-S109 in embodiment one It causes, details are not described herein again.
S304, calculate node are according to storage resource response message, and using the first data link processes read-write requests, this first The link that data link is established for the first memory node and calculate node, first memory node be in memory node cluster with meter The corresponding memory node currently in use of the first volume that operator node uses.
It, should after calculate node receives the storage resource response message that the response storage resource request of memory node cluster is sent Calculate node can be read and write according to storage resource response message by multipath Systematic selection using the first data link processes It requests.Wherein, the first memory node is used to be corresponding with the first volume that calculate node uses in memory node cluster Memory node.
S305, when using the first data link processes read-write requests, calculate node detects the company of first data link Connect state.
The description during S305 in the embodiment of the present invention is consistent with the description of the S101 in embodiment one, herein no longer It repeats.
If S306, detecting that the first data link is in off-state, calculate node, which is sent, updates the data request to second Memory node, this updates the data request for the current meta data information for asking the download first volume, and second memory node is deposits It stores up in node cluster in addition to the first memory node, slack storage node corresponding with the first volume that calculate node uses, this first Roll up the storage unit used for calculate node.
The description during S306 in the embodiment of the present invention is consistent with the description of the S102 in embodiment one, herein no longer It repeats.
S307, the response of the second memory node update the data request, and the current meta data information of the first volume is downloaded in request.
The description during S307 in the embodiment of the present invention is consistent with the description of the S202 in embodiment two, herein no longer It repeats.
If the current meta data information of S308, the first volume is downloaded successfully, the second memory node, which is sent, to be updated successfully more New response message is to calculate node.
The description during S308 in the embodiment of the present invention is consistent with the description of the S203 in embodiment two, herein no longer It repeats.
If to be updated successfully, calculate node, which is sent, handovers request to the second memory node for S309, update response message, should Handover request is used to the link for handling read-write requests being switched to the second data link from the first data link, second data-link The link that road is established for the second memory node with calculate node.
The description during S309 in the embodiment of the present invention is consistent with the description of the S104 in embodiment one, herein no longer It repeats.
S310, the second memory node use the second data link processes read-write requests according to handover request.
The description during S310 in the embodiment of the present invention is consistent with the description of the S204 in embodiment two, herein no longer It repeats.
Further, as shown in Figure 10, after S307, a kind of information processing method provided in an embodiment of the present invention also wraps It includes：S311.It is as follows：
If the current meta data information failed download of S311, the first volume, the second memory node sends update failure more New response message is to calculate node.
The description during S311 in the embodiment of the present invention is consistent with the description of the S205 in embodiment two, herein no longer It repeats.
Further, as shown in figure 11, after S308, a kind of information processing method provided in an embodiment of the present invention also wraps It includes：S312.It is as follows：
If S312, update response message fail for update, calculate node, which forbids sending, handovers request to the second storage section Point.
The description during S307 in the embodiment of the present invention is consistent with the description of the S105 in embodiment one, herein no longer It repeats.
Further, as shown in figure 12, after S305, a kind of information processing method provided in an embodiment of the present invention also wraps It includes：S313.It is as follows：
If S313, when detecting that the first data link is in connected state, calculate node is continuing with first data-link Road handles read-write requests.
The description during S313 in the embodiment of the present invention is consistent with the description of the S106 in embodiment one, herein no longer It repeats.
An embodiment of the present invention provides a kind of information processing method, when using the first data link processes read-write requests, The connection status of the first data link is detected, the first data link is the link that the first memory node is established with calculate node, the One memory node is memory node currently in use corresponding with the first volume that calculate node uses in memory node cluster；If inspection It measures the first data link and is in off-state, then send and update the data request to the second memory node, update the data request and be used for Request download the first volume current meta data information, the second memory node be memory node cluster in addition to the first memory node, Slack storage node corresponding with the first volume that calculate node uses；It receives the response of the second memory node and updates the data request transmission Update response message；If update response message is is updated successfully, transmission handovers request to the second memory node, handover request For the link for handling read-write requests to be switched to the second data link from the first data link, the second data link is deposited for second Store up the link that node is established with calculate node.Using above-mentioned technic relization scheme, since calculate node is used detecting A data link disconnect i.e. break down when, the calculate node can first inform will be switched to second storage section Point first carries out the update of the metadata information of the first volume, therefore, the metadata information of the first volume is had updated in the second memory node Afterwards, calculate node can be written and read the processing of request using the first volume in the corresponding data link of the second memory node , the normal processing of read-write requests is ensured that in this way, and improves the accuracy rate of read-write requests.
Example IV
As shown in figure 13, an embodiment of the present invention provides a kind of calculate node 1, corresponding to the information processing of calculate node side Method, the calculate node 1 can include：
Detection unit 10, for when using the first data link processes read-write requests, detecting first data link Connection status, the link that first data link is established for the first memory node and the calculate node, described first deposits Node is stored up as memory node currently in use corresponding with the first volume that the calculate node uses in memory node cluster.
First transmitting element 11, if for detecting that first data link is in off-state, transmission updates the data Request is described to update the data request for asking the current meta data information of the download first volume, institute to the second memory node The second memory node is stated as the institute that in addition to first memory node, is used with the calculate node in the memory node cluster State the corresponding slack storage node of the first volume.
First receiving unit 12, for receiving the update that request is updated the data described in the second memory node response and is sent Response message.
First transmitting element 11 if being additionally operable to the update response message to be updated successfully, sends handover request To second memory node, the link that the handover request is used to handle read-write requests switches from first data link The link established for the second data link, second data link for second memory node with the calculate node.
Optionally, as shown in figure 14, the calculate node 1 further includes：Forbid unit 13.
It is described to forbid unit, for described receive request transmission is updated the data described in the second memory node response more Newly after response message, if the update response message fails for update, transmission is forbidden to handover request to second storage Node.
Optionally, as shown in figure 15, the calculate node 1 further includes：Read-write cell 14.
The read-write cell 14, after the connection status for detecting first data link, if detecting institute When stating the first data link and being in connected state, then the first data link processes read-write requests are continuing with.
Optionally, first receiving unit 12 is additionally operable to before the first data link processes read-write requests of the use, Receive the first read write command.
First transmitting element 11 is additionally operable to respond first read write command, sends storage resource and asks to described Memory node cluster.
First receiving unit 12 is additionally operable to receive the memory node cluster response storage resource request transmission Storage resource response message, carry the corresponding institute of the first volume, the first volume in the storage resource response message State the first memory node and second memory node.
As shown in figure 16, in practical applications, above-mentioned detection unit 10, forbid unit 13 and read-write cell 14 can be by being located at Processor 15 in calculate node realizes, specially central processing unit (CPU), microprocessor (MPU), digital signal processor (DSP) or realizations, first receiving unit 12 such as field programmable gate array (FPGA) are realized by receiver 16, the first transmitting element 11 are realized by transmitter 17, and the receiver 16 and transmitter 17 can be connect by system bus 18 with processor 15.
Optionally, calculate node provided in an embodiment of the present invention can be server, but without the service of storage region Device.
An embodiment of the present invention provides a kind of calculate nodes, when using the first data link processes read-write requests, detection The connection status of first data link, the link that the first data link is established for the first memory node with calculate node, first deposits Node is stored up as memory node currently in use corresponding with the first volume that calculate node uses in memory node cluster；If it detects First data link is in off-state, then sends and update the data request to the second memory node, update the data request for asking The current meta data information of the first volume is downloaded, the second memory node is in memory node cluster in addition to the first memory node, with meter The corresponding slack storage node of the first volume that operator node uses；It receives the response of the second memory node and updates the data request transmission more New response message；If update response message is is updated successfully, transmission handovers request to the second memory node, and handover request is used for The link for handling read-write requests is switched to the second data link from the first data link, the second data link is saved for the second storage The link that point is established with calculate node.Using above-mentioned technic relization scheme, since calculate node is detecting currently in use one When data link disconnects i.e. failure, which can first inform that the second memory node that will be switched to is first The update of the metadata information of the first volume is carried out, therefore, after the second memory node has updated the metadata information of the first volume, meter Operator node can be written and read the processing of request using the first volume in the corresponding data link of the second memory node, this Sample ensures that the normal processing of read-write requests, and improves the accuracy rate of read-write requests.
As shown in figure 17, an embodiment of the present invention provides a kind of second memory node 2, corresponding to the second storage-node side Information processing method, second memory node 2 can include：
Second receiving unit 20, for receive calculate node transmission update the data request.
Download unit 21, for responding the current meta data information for updating the data request, asking to download the first volume, institute It is the storage unit that the calculate node uses to state the first volume.
Second transmitting element 22, if the current meta data information for the first volume is downloaded successfully, transmission is updated to The update response message of work(is to the calculate node.
Second receiving unit 20 is additionally operable to receive the handover request that the calculate node is sent, according to the switching Using the second data link processes read-write requests, second data link is second memory node and the calculating for request The link that node is established.
Optionally, second transmitting element 22 is additionally operable to update the data request described in the response, and request downloads first After the current meta data information of volume, if the current meta data information failed download of the first volume, update failure is sent Response message is updated to the calculate node.
As shown in figure 18, in practical applications, above-mentioned download unit 21 can be by being located at the processor on the second memory node 23 realize, specially central processing unit (CPU), microprocessor (MPU), digital signal processor (DSP) or field programmable gate The realizations such as array (FPGA), the second receiving unit 20 are realized that the second transmitting element 22 is realized by transmitter 25 by receiver 24, on The second memory node is stated to further include：Storage medium 26 realizes that the storage medium 26, receiver 24 and transmitter 25, which can pass through, is System bus 27 is connect with processor 23, wherein, for storage medium 26 for storing executable program code, which includes meter Calculation machine operational order, storage medium 26 may include high-speed RAM memory, it is also possible to nonvolatile memory is further included, for example, At least one magnetic disk storage.
Optionally, the second memory node in the embodiment of the present invention can be block storage server, and the embodiment of the present invention is not It is restricted.
An embodiment of the present invention provides a kind of second memory nodes, and receive calculate node transmission updates the data request；It rings Request should be updated the data, the current meta data information of the first volume is downloaded in request, and the first volume is the storage that the calculate node uses Unit；If the current meta data information of the first volume, which is downloaded, successfully, sends the update response message that is updated successfully to calculate node； The handover request that calculate node is sent is received, the second data link processes read-write requests, the second data are used according to handover request The link that link is established for second memory node with calculate node.Using above-mentioned technic relization scheme, due to calculate node When detecting that a data link currently in use disconnects i.e. failure, which can first inform and will cut The second memory node changed to first carries out the update of the metadata information of the first volume, therefore, is had updated in the second memory node After a roll of metadata information, calculate node can involve in the corresponding data link of the second memory node using first The processing of row read-write requests, ensures that the normal processing of read-write requests in this way, and improves the accuracy rate of read-write requests.
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, system or computer program Product.Therefore, the shape of the embodiment in terms of hardware embodiment, software implementation or combination software and hardware can be used in the present invention Formula.Moreover, the present invention can be used can use storage in one or more computers for wherein including computer usable program code The form of computer program product that medium is implemented on (including but not limited to magnetic disk storage and optical memory etc.).
The present invention be with reference to according to the method for the embodiment of the present invention, the flow of equipment (system) and computer program product Figure and/or block diagram describe.It should be understood that it can be realized by computer program instructions every first-class in flowchart and/or the block diagram The combination of flow and/or box in journey and/or box and flowchart and/or the block diagram.These computer programs can be provided The processor of all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce A raw machine so that the instruction performed by computer or the processor of other programmable data processing devices is generated for real The device of function specified in present one flow of flow chart or one box of multiple flows and/or block diagram or multiple boxes.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy Determine in the computer-readable memory that mode works so that the instruction generation being stored in the computer-readable memory includes referring to Enable the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one box of block diagram or The function of being specified in multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted Series of operation steps are performed on calculation machine or other programmable devices to generate computer implemented processing, so as in computer or The instruction offer performed on other programmable devices is used to implement in one flow of flow chart or multiple flows and/or block diagram one The step of function of being specified in a box or multiple boxes.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.

Claims (12)

1. a kind of information processing method, which is characterized in that applied in calculate node, including：
When using the first data link processes read-write requests, the connection status of detection first data link, described first The link that data link is established for the first memory node with the calculate node, first memory node are memory node cluster In memory node currently in use corresponding with the first volume that the calculate node uses；
If detecting, first data link is in off-state, and transmission updates the data request to the second memory node, described Request is updated the data for asking the current meta data information of the download first volume, second memory node is the storage In node cluster in addition to first memory node, slack storage section corresponding with the first volume that the calculate node uses Point；
It receives and updates the data the update response message that request is sent described in the second memory node response；
If the update response message is is updated successfully, transmission handovers request to second memory node, the switching please It asks the link for handling read-write requests being switched to the second data link, second data-link from first data link The link that road is established for second memory node with the calculate node.
2. according to the method described in claim 1, it is characterized in that, described receive the second memory node response update After the update response message that request of data is sent, the method further includes：
If the update response message fails for update, transmission is forbidden to handover request to second memory node.
3. according to the method described in claim 1, it is characterized in that, the connection status of the detection first data link it Afterwards, the method further includes：
If detect that first data link is in connected state, being continuing with the first data link processes read-write please It asks.
4. according to the method described in claim 1, it is characterized in that, it is described using the first data link processes read-write requests it Before, the method further includes：
Receive the first read write command；
First read write command is responded, storage resource is sent and asks to the memory node cluster；
It receives the memory node cluster and responds the storage resource response message that the storage resource request is sent, the storage money Corresponding first memory node of the first volume, the first volume and the second storage section are carried in the response message of source Point.
5. a kind of information processing method, which is characterized in that applied in the second memory node, including：
Receive calculate node transmission updates the data request；
Request is updated the data described in response, the current meta data information of the first volume is downloaded in request, and the first volume is the calculating The storage unit that node uses；
If the current meta data information of the first volume, which is downloaded, successfully, sends the update response message that is updated successfully to the meter Operator node；
The handover request that the calculate node is sent is received, it please using the read-write of the second data link processes according to the handover request It asks, the link that second data link is established for second memory node with the calculate node.
6. according to the method described in claim 5, it is characterized in that, update the data request described in the response, request downloads the After a roll of current meta data information, the method further includes：
If the current meta data information failed download of the first volume, the update response message of update failure is sent to the meter Operator node.
7. a kind of calculate node, which is characterized in that including：
Detection unit, for when using the first data link processes read-write requests, detecting the connection of first data link State, the link that first data link is established for the first memory node with the calculate node, first memory node For memory node currently in use corresponding with the first volume that the calculate node uses in memory node cluster；
First transmitting element, if for detecting that first data link is in off-state, transmission updates the data request extremely Second memory node, the current meta data information for updating the data request and the first volume being downloaded for request, described second Memory node be the memory node cluster in addition to first memory node, used with the calculate node described first Roll up corresponding slack storage node；
First receiving unit disappears for receiving to update the data the update for asking to send described in the second memory node response and respond Breath；
First transmitting element, if being additionally operable to the update response message to be updated successfully, transmission handovers request to described Second memory node, the handover request are used to the link for handling read-write requests being switched to second from first data link Data link, the link that second data link is established for second memory node with the calculate node.
8. node according to claim 7, which is characterized in that the calculate node further includes：Forbid unit；
It is described to forbid unit, receive the update sound that request is updated the data described in the second memory node response and is sent for described After answering message, if the update response message fails for update, transmission is forbidden to handover request to second memory node.
9. node according to claim 7, which is characterized in that the calculate node further includes：Read-write cell；
The read-write cell, after the connection status for detecting first data link, if detecting described first When data link is in connected state, then the first data link processes read-write requests are continuing with.
10. node according to claim 7, which is characterized in that
First receiving unit is additionally operable to before the first data link processes read-write requests of the use, receives the first read-write Instruction；
First transmitting element is additionally operable to respond first read write command, sends storage resource and asks to the storage section Point cluster；
First receiving unit is additionally operable to receive the storage that the memory node cluster responds the storage resource request transmission Resource response message carries the first volume, the first volume corresponding described first in the storage resource response message Memory node and second memory node.
11. a kind of second memory node, which is characterized in that including：
Second receiving unit, for receive calculate node transmission update the data request；
Download unit, for respond it is described update the data request, the current meta data information of the request download first volume, described first Roll up the storage unit used for the calculate node；
Second transmitting element, if the current meta data information for the first volume is downloaded successfully, transmission is updated successfully more New response message is to the calculate node；
Second receiving unit is additionally operable to receive the handover request that the calculate node is sent, be made according to the handover request With the second data link processes read-write requests, second data link is built for second memory node with the calculate node Vertical link.
12. node according to claim 11, which is characterized in that
Second transmitting element is additionally operable to update the data request described in the response, and current first number of the first volume is downloaded in request It is believed that after breath, if the current meta data information failed download of the first volume, the update response message of update failure is sent To the calculate node.
CN201611270149.7A 2016-12-30 2016-12-30 Information processing method, computing node and storage node Active CN108268210B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611270149.7A CN108268210B (en) 2016-12-30 2016-12-30 Information processing method, computing node and storage node

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611270149.7A CN108268210B (en) 2016-12-30 2016-12-30 Information processing method, computing node and storage node

Publications (2)

Publication Number Publication Date
CN108268210A true CN108268210A (en) 2018-07-10
CN108268210B CN108268210B (en) 2022-03-08

Family

ID=62770500

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611270149.7A Active CN108268210B (en) 2016-12-30 2016-12-30 Information processing method, computing node and storage node

Country Status (1)

Country Link
CN (1) CN108268210B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109358812A (en) * 2018-10-09 2019-02-19 郑州云海信息技术有限公司 Processing method, device and the relevant device of I/O Request in a kind of group system
CN112307113A (en) * 2019-07-29 2021-02-02 中兴通讯股份有限公司 Service request message sending method and distributed database architecture

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7096333B2 (en) * 2002-07-18 2006-08-22 International Business Machines Corporation Limited concurrent host access in a logical volume management data storage environment
CN102868754A (en) * 2012-09-26 2013-01-09 北京联创信安科技有限公司 High-availability method, node device and system for achieving cluster storage
CN104461792A (en) * 2014-12-03 2015-03-25 浪潮集团有限公司 HA method for solving NAMENODE single-point fault of HADOOP distributed file system
CN105069160A (en) * 2015-08-26 2015-11-18 国家电网公司 Autonomous controllable database based high-availability method and architecture
CN106126652A (en) * 2016-06-24 2016-11-16 武汉斗鱼网络科技有限公司 Mishap Database changing method and system for distributed experiment & measurement system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7096333B2 (en) * 2002-07-18 2006-08-22 International Business Machines Corporation Limited concurrent host access in a logical volume management data storage environment
CN102868754A (en) * 2012-09-26 2013-01-09 北京联创信安科技有限公司 High-availability method, node device and system for achieving cluster storage
CN104461792A (en) * 2014-12-03 2015-03-25 浪潮集团有限公司 HA method for solving NAMENODE single-point fault of HADOOP distributed file system
CN105069160A (en) * 2015-08-26 2015-11-18 国家电网公司 Autonomous controllable database based high-availability method and architecture
CN106126652A (en) * 2016-06-24 2016-11-16 武汉斗鱼网络科技有限公司 Mishap Database changing method and system for distributed experiment & measurement system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109358812A (en) * 2018-10-09 2019-02-19 郑州云海信息技术有限公司 Processing method, device and the relevant device of I/O Request in a kind of group system
CN112307113A (en) * 2019-07-29 2021-02-02 中兴通讯股份有限公司 Service request message sending method and distributed database architecture

Also Published As

Publication number Publication date
CN108268210B (en) 2022-03-08

Similar Documents

Publication Publication Date Title
US8825851B2 (en) Management of a virtual machine in a storage area network environment
US9619311B2 (en) Error identification and handling in storage area networks
JP4107083B2 (en) High-availability disk controller, its failure handling method, and high-availability disk subsystem
US8806007B2 (en) Inter-node communication scheme for node status sharing
CN105262633B (en) A kind of application redundancy method and application redundancy system
CN102469025A (en) Multipath switching over multiple storage systems
CN101799742A (en) Methods and apparatus for migrating thin provisioning volumes between storage systems
WO2011151861A1 (en) Storage system having sas as its backend communication standard
JP2008009497A (en) Storage subsystem, host computer, storage system, method of distributing load, and load distribution program
CN102629225A (en) Dual-controller disk array, storage system and data storage path switching method
EP2723017A1 (en) Method, apparatus and system for implementing distributed auto-incrementing counting
JP2006260355A (en) Storage system and storage control method
CN105786405A (en) Online upgrading method, device and system
JP4596889B2 (en) Storage system management method
CN109101357A (en) A kind of detection method and device of OSD failure
CN105446657A (en) Method for monitoring RAID card
CN108459824A (en) A kind of data modification write method and device
US20140052845A1 (en) Discovery of storage area network devices for a virtual machine
CN108268210A (en) A kind of information processing method, calculate node and memory node
JP2005251188A (en) Redundant path control apparatus and redundant path control method
CN107888405B (en) Management apparatus and information processing system
CN103647710A (en) Path configuration method and device
CN104184603A (en) User status accounting method, device and system
US20130290772A1 (en) Sequence indicator for command communicated to a sequential access storage device
JP4357433B2 (en) Storage system

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
GR01 Patent grant
GR01 Patent grant