WO2017101016A1

WO2017101016A1 - Method and apparatus for synchronizing service request of storage node

Info

Publication number: WO2017101016A1
Application number: PCT/CN2015/097431
Authority: WO
Inventors: 于传帅; 张程伟
Original assignee: 华为技术有限公司
Priority date: 2015-12-15
Filing date: 2015-12-15
Publication date: 2017-06-22
Also published as: CN107710165A; CN107710165B

Abstract

The embodiments of the present invention relate to a method and apparatus for synchronizing a service request of a storage node. The method comprises: when a first storage node has a fault within t₁ to t₂ while executing a plurality of service requests, determining a set of service requests, which are recorded in a first log, and of which a starting execution moment is within (t₁-2dt) to (t₂+2dt), as a first service request set, and a set of service requests, which are recorded in a second log, and of which a starting execution moment is within (t₁-dt) to (t₂+dt), as a second service request set, wherein dt is the maximum time span for executing the service requests; and determining that a service request belonging to the second service request set but not belonging to the first service request set is a service request lost by the first storage node during the occurrence of the fault so as to facilitate the first storage node executing the lost service request. By means of the method and apparatus for synchronizing a service request of a storage node in the embodiments of the invention, a lost service request can be obtained accurately, and service requests of a normal storage node and a faulty storage node can be synchronized, so that the throughput of a storage system in executing the service requests is improved.

Description

Method and apparatus for storing node synchronization service requests

Technical field

The present invention relates to the field of communications, and in particular, to a method and apparatus for storing node synchronization service requests.

Background technique

In a large-scale storage system, storage node failure is the normal state. When the storage node fails, it is necessary to ensure that the data is not lost. This is high reliability, and the system continues to provide services. This is high availability. Storage vendors generally use multiple copies. Or Erasure code (EC) technology, such as copying multiple copies (multiple copies) of data, storing multiple copies to provide high reliability and high availability.

It is assumed that the storage system is still delivering services normally during the storage node failure. In this case, the faulty storage node loses part of the service request, and the faulty storage node needs to synchronize with the service request of the normal storage node.

In the existing strong consistent storage system storage, a State Machine (Replication State Machine (RSM) mechanism is generally used to provide synchronization technology. The basic idea of the RSM mechanism is that each storage node is completely synchronized. For example, if there are four service requests, put(key1), put(key2), del(key1), and put(key3), each storage node must execute put ( Key1), then execute put(key2), then execute del(key1), then execute put(key3), and record the log in the same order at the same time. If some service requests are lost during the storage node failure, these service requests are in the log (log ) must be close and continuous, for example, the last two service requests del (key1), put (key3), after the fault storage node restarts, it is easy to find the lost service request by comparing the log, fill this Two business requests can be synchronized.

However, in order to efficiently recover the lost service request of the storage node in order to efficiently identify the fault, the service request must be executed sequentially, and the serial number of the service request needs to be allocated and maintained, which greatly limits the throughput of the storage system to perform the service request. .

Summary of the invention

The present invention provides a method and apparatus for storing a node synchronization service request, which can accurately obtain a lost service request, synchronize a service request of a normal storage node and a faulty storage node, and improve the throughput of the storage system to perform a service request.

In a first aspect, a method for storing a node synchronization service request is provided, the method being applied to a storage system including a first storage node and a second storage node, the first storage node being configured to execute in a first order The service requests and stores a first log, where the first log is used to record each service request that the first storage node has executed in the multiple service requests and the start execution of each service request that the first storage node has performed At a time, the second storage node is configured to execute the multiple service requests in a second order and store a second log, where the second log is used to record each service that the second storage node has executed in the multiple service requests. The request and the start execution time of each service request that the second storage node has executed, the method comprising: when the first storage node fails within t ₁ to t ₂ of executing the plurality of service requests, the service request is synchronized The device determines a first service request set and a second service request set, where the first service request set is a start execution time recorded in the first log at t ₁ -2dt to t ₂ + a set of service requests within 2dt, the second set of service requests being a set of service requests recorded in the second log from t ₁ -dt to t ₂ +dt, the dt being greater than or equal to the first a difference between a time when the storage node starts to execute the target service request and a time when the second storage node starts to execute the target service request, where the target service request is the service in the first service request set or the second service request set a request that the service request synchronization device determines that the service request belonging to the second service request set and does not belong to the first service request set is a service request lost by the first storage node during a failure; the service request synchronization device indicates the first The second storage node sends the lost service request to the first storage node.

Therefore, in the method for storing a node synchronization service request in the embodiment of the present invention, two storage nodes in the storage system may not perform the service request in the order in which the service request is delivered, and may find different time when one of the failures occurs. A service request that is not a service request recorded by the failed storage node but belongs to the normal storage node record is a lost service request, so that different storage nodes can perform service requests concurrently or out of order without special allocation and The serial number of the service request is maintained, and the lost service request can be accurately obtained from the normal storage node, the service request synchronization is completed, and the throughput of the storage system executing the service request is improved.

With reference to the first aspect, in an implementation manner of the first aspect, the service request synchronization device determines that the service request that belongs to the second service request set and does not belong to the first service request set is that the first storage node is faulty. The service request lost during the period includes: the service request synchronization device respectively sorts the first service request set and the service request in the second service request set according to a key of the service request; the service request synchronization device is compared according to the sorted result ratio Receiving, by the first service request set and the service request in the second service request set, the second service request set And the service request that does not belong to the first service request set is the lost service request, so that the traversal action can be avoided, and the lost service request can be searched more quickly.

In conjunction with the first aspect and the foregoing implementation manner, in another implementation manner of the first aspect, the service request synchronization device determines, that the service request that belongs to the second service request set and does not belong to the first service request set is the first a service request lost by a storage node during a failure, the: the service request synchronization device determines a flag bitmap of the first service request set according to a key of each service request in the first service request set; the service request is synchronized The device determines, according to a key of each service request in the second service request set, a flag bitmap of the second service request set; the service request synchronization device compares the flag bit corresponding to the first service request set with the second service The corresponding flag bit of the request set is obtained, and the service request belonging to the second service request set and not belonging to the first service request set is the lost service request, so that the lost service request can be determined more intuitively.

In conjunction with the first aspect and the foregoing implementation manner, in another implementation manner of the first aspect, the service request synchronization device determines, that the service request that belongs to the second service request set and does not belong to the first service request set is the first a service request lost by a storage node during a failure, comprising: the service request synchronization device determining the first service request set according to a Bloom filter bloom filter algorithm and a key of each service request in the first service request set Bitmap; the service request synchronization device determines a bitmap of the second service request set according to the bloom filter algorithm and a key of each service request in the second service request set; the service request synchronization device compares the first service The bitmap of the request set and the bitmap of the second service request set determine that the service request belonging to the second service request set and not belonging to the first service request set is the lost service request.

In combination with the first aspect and the foregoing implementation manner, in another implementation manner of the first aspect, the first sequence is different from the second sequence.

In conjunction with the first aspect and the foregoing implementation manner, in another implementation manner of the first aspect, the service request synchronization apparatus, the second storage node, instructs the second storage node to send the lost service request to the first storage node, including: the service The request synchronization device receives the lost service request sent by the second storage node; the service request synchronization device sends the lost service request to the first storage node.

In a second aspect, an apparatus for storing a node synchronization service request is provided for performing the method of any of the first aspect or the first aspect of the first aspect. In particular, the apparatus comprises means for performing the method of any of the above-described first aspect or any of the possible implementations of the first aspect.

A third aspect provides an apparatus for storing a node synchronization service request, including: storing a unit and a processor for storing instructions for executing instructions stored by the memory, and when the processor executes the instructions stored by the memory, the executing causes the processor to perform the first aspect or the first A method in any possible implementation of the aspect.

In a fourth aspect, a computer readable medium is provided for storing a computer program comprising instructions for performing the method of the first aspect or any of the possible implementations of the first aspect.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings to be used in the embodiments of the present invention will be briefly described below.

1 is a schematic diagram of a storage cluster in accordance with an embodiment of the present invention.

2 is a schematic flowchart of a method for storing a node synchronization service request according to an embodiment of the present invention.

3 is a schematic diagram of a method of determining a set of service requests in accordance with an embodiment of the present invention.

4 is another schematic diagram of a method of determining a set of service requests in accordance with an embodiment of the present invention.

FIG. 5 is still another schematic diagram of a method of determining a service request set according to an embodiment of the present invention.

6 is a schematic block diagram of an apparatus for storing a node synchronization service request, in accordance with an embodiment of the present invention.

FIG. 7 is a schematic block diagram of an apparatus for storing a node synchronization service request according to another embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention will be clearly described in conjunction with the drawings in the embodiments of the present invention. It is obvious that the described embodiments are a part of the embodiments of the present invention, and not all embodiments.

As shown in FIG. 1 , FIG. 1 is a physical structure of a storage cluster, and the key-value storage node may be a storage cluster that provides a key-value interface as shown in FIG. 1 , and the storage cluster may include multiple Ethernets (Ethernet). A protocol interface hard disk, also known as an IP disk. For example, the storage cluster shown in FIG. 1 includes two IP disks, or may include more than two IP disks.

In addition, the key-value storage node may also be a single IP disk that provides a key-value interface; the key-value storage node may also be a key-value interface deployed in a cluster or a single disk. Virtual device.

The embodiment of the present invention describes a method for storing a node synchronization service request. The method is applied to a storage system, and the storage system may include two key-value storage nodes as described above. In the storage system, the service request can be executed concurrently and out of order. Each key-value storage node stores the key-value data of the storage node and the log corresponding to the storage node, that is, the storage system Save two key-value data and save the logs corresponding to the two storage nodes. After the storage node 1 in the storage system is restarted, the service request of the storage node 1 is in the log (log) 1 corresponding to the storage node because the service request in the storage node is executable in an out-of-order manner. The present invention describes a method for storing a node synchronization service request, which can accurately obtain a service request that is lost by the failed storage node, and is not immediately and continuously. Completing the synchronization of service requests improves the throughput of the storage system to perform business requests.

2 shows a schematic flowchart of a method 100 for storing a node synchronization service request, which is applied to a storage system including a first storage node and a second storage node, the first, in accordance with an embodiment of the present invention. The storage node is configured to execute a plurality of service requests in a first order, where the first storage node stores a first log, where the first log is used to record each service request that the first storage node has executed in the multiple service requests And the start execution time of the each service request, the second storage node is configured to execute the plurality of service requests in a second order, the second storage node stores a first log, and the second log is used to record in the multiple In the service request, each service request that the second storage node has performed and the start execution time of the each service request, the method 100 may be performed by a service request synchronization device, and the device may be a virtual module, and the device is located in the storage Any one of the storage nodes in the system, such as the first storage node or the second storage node, or other storage node, is not limited to the embodiment of the present invention. herein. As shown in FIG. 2, the method 100 includes:

S110, when the first storage node in the plurality of service execution request t ₁ to t ₂ when a failure occurs, a service request synchronizing means determines a first set and a second service request service request set, the first set for the service request a set of service requests recorded in the first log at t ₁ -2dt to t ₂ +2dt, the second service request set being the start execution time recorded in the second log at t ₁ -dt to a set of service requests within t ₂ +dt, the dt being greater than or equal to a difference between a time at which the first storage node begins to execute the target service request and a time at which the second storage node begins to execute the target service request, the target The service request is a service request in the first service request set or the second service request set;

S120. The service request synchronization device determines that the service request that belongs to the second service request set and does not belong to the first service request set is a service request that is lost during the fault of the first storage node.

S130. The service request synchronization device instructs the second storage node to send the lost service request to the first storage node.

Specifically, the first storage node executes a plurality of service requests in a first order, and records, by the first log, each service request that has been executed and a corresponding start execution time, and the second storage node also executes the plurality of times according to the second sequence. service request, has been performed by the second log record for each service request and corresponding start execution time, when the time t ₂ to t _1, a first storage node fails, it is determined that the first service in the first log a request set, the first service request set includes a service request executed in a first log at a time t ₁ -2dt to t ₂ +2dt; and a second service request set is also determined in the second log, the second service The request set includes a service request executed in a second log at a time t ₁ -dt to t ₂ +dt, wherein the dt may be referred to as a maximum time span, and dt is greater than or equal to the first storage node to start performing a target service request The difference between the time of the second storage node and the time when the second storage node starts to execute the target service request, where the target service request is any one of multiple service requests sent by the system Requesting, optionally, selecting, for the target service, one of all service requests belonging to the first service request set or the second service request set, and all service requests included in the first service request set and the second service request set Can meet the range of the maximum time span dt. And determining, by the service request that belongs to the second service request set but not belonging to the first service request set, the service request that is lost by the first storage node during the fault period, that is, the lost service request is not executed by the first storage node due to the fault. Service request so that the first storage node performs the lost service request.

Therefore, in the method for storing a node synchronization service request in the embodiment of the present invention, two storage nodes in the storage system may not perform the service request in the order in which the service request is delivered, and when one of the failures occurs, the search may be different. The service request that is not recorded by the failed storage node during the time period but belongs to the normal storage node record is a lost service request, so that different storage nodes can execute the service request concurrently or out of order without special allocation. And maintaining the serial number of the service request, and also being able to accurately obtain the lost service request from the normal storage node, complete the service request synchronization, and improve the throughput of the storage system to perform the service request.

In the embodiment of the present invention, for a plurality of service requests sent by the storage system, the first storage node may perform the multiple service requests in a first order, and the second storage node performs the second sequence in the second order. For multiple service requests, the first sequence and the second sequence may be the same or different. The order of executing multiple service requests may be inconsistent with the delivery order, and multiple service requests may be executed concurrently or in an out-of-order manner. The serial number of the service request is assigned and maintained, which limits the service request throughput. In addition, for each storage node to record the executed service request, the first log records the service request that has been executed by the first storage node and the start execution time of the service request, and records the executed service of the second storage node by using the second log. The request and the start of execution of the business request.

In the embodiment of the present invention, the first storage node or the second storage node does not limit the execution order of the service request, but may limit the execution time of the service request. If the storage node does not perform the service when the preset time is exceeded. The request, that is, the storage system does not receive the response message returned by the storage node after executing the service request within a certain period of time, the system may re-issue the service request, and the storage node should re-execute the service request. Optionally, the preset time may be set according to the actual application. After the time is set in advance, if the response message of the storage node is not received, the system may re-issue the service request. The storage node can re-execute the service request.

In the embodiment of the present invention, a maximum time span may be defined, that is, when any one of the service requests, for example, the target service request is executed at time t, appears in a log, that is, is executed by a storage node, then at t-dt During the time period t+dt, the target service request must also appear in another log, and the other storage node also starts to execute, where dt is the maximum time span. For example, at time t, the target service request appears in the first log, that is, the first storage node starts execution, then the service request must appear in the second time during the period from t-dt to t+dt. The log is executed by the second storage node, that is, the dt satisfies a time greater than or equal to a time when the first storage node starts executing the target service request and a time when the second storage node starts executing the target service request. Difference. Optionally, the maximum time span dt may be preset according to actual conditions, and each service request performed by each storage node meets the requirement of the maximum time span. Optionally, the maximum time span dt may also be calculated after the fault occurs, and the maximum time span of the service request in a period of time is calculated. The embodiment of the present invention is not limited thereto.

In S110, when the first storage node in the plurality of service execution request, _{t. 1} to t in the event of a failure within, the first service time of the first request set ₂ is determined according to the log, the log is determined in accordance with a second a second service request set, where the first service request set is a set of executed service requests recorded in the first log from t ₁ -2dt to t ₂ +2dt, and the second service request set is the second log a set of executed service requests recorded at times t ₁ -dt to t ₂ +dt, wherein the dt is a maximum time span, and the dt may be based on the services included in the first service request set and the second service request set The request is obtained by the calculation; in S120, the service request that belongs to the second service request set and does not belong to the first service request set is a service request that is lost during the failure of the first storage node, so as to restore the normal A storage node can re-execute the lost service request during the failure, thereby implementing service request synchronization and improving the throughput of the storage system to perform service requests.

In the embodiment of the present invention, the first storage node and the second storage node meet the requirement of the maximum time span when performing the service request. When any storage node fails, for example, as shown in FIG. 3, the first storage node is at t. _The fault occurs in the time from ₁ to t ₂ , that is, the first storage fails during the period A in Figure 3, and the service request lost by the first storage node during this period must belong to the service request in the B period in Figure 3, That is, the service request executed by the second storage node in the time t ₁ -dt to t ₂ +dt, that is, in the service request recorded in the second log from t ₁ -dt to t ₂ +dt, these will be The service request recorded in the segment time is determined as the second service request set; and the lost service request does not belong to the C segment time, that is, the service request executed by the first storage node in the time t ₁ -2dt to t ₂ +2dt, that is, does not appear in the first log t ₁ -2dt to t ₂ + 2dt service request within the time recorded, the service requests the time period is determined to record a first set of service request. Therefore, the service request lost during the failure of the first storage node is determined, that is, the service request belonging to the A-segment time is determined, that is, the service request that belongs to the second service request set but does not belong to the first service request set is the first storage. A business request that the node is missing.

Alternatively, as an embodiment, when a fault occurs in the first storage node t ₁ to time t _2, the failure determining device of the first storage node is lost service request is determined to belong to a second set of the service request, but The service request that is not in the first service request set, the first service request set is a service request that is performed by the first storage node in the time t ₁ -2dt to t ₂ +2dt, and is recorded in t ₁ -2dt in the first log to t ₂ + 2dt time; the second service request is combined with the service request executed by the second storage node in the time t ₁ -dt to t ₂ +dt, recorded in the second log t ₁ -dt to t ₂ +dt In time. Specifically, the lost service request may be determined by sorting the service request in the first service request set and the second service request according to the key. For example, as shown in FIG. 4, A, B, and C respectively include a plurality of service requests, where the service request is an input key (for example, put(key1), and the serial number 1 corresponds to a service request key. In addition, A corresponds to a lost service request, B corresponds to a second service request set, C corresponds to the first service request set, and determines a service request that belongs to the second service request set but does not belong to the first service request set, that is, belongs to B. However, if the service request does not belong to C, the service requests in B and C can be sorted according to the key. The sort result is shown in Figure 4. According to the order of the keys, the service requests belonging to B but not belonging to C are respectively Key2, key4, key6, and key7 are consistent with those in A, indicating that the lost service request can be accurately determined in this way. By using this sorting method, you can avoid traversing each service request recorded in the log. For example, for key6 in B, if you find that it belongs to C, you can find that only C is included in C, and then key8 is used. It is determined that key6 is missing data, and it can also be determined that key7 must also be lost data, without having to judge key7 again, to avoid traversing the operation of each service request in B.

Optionally, as an embodiment, the lost service request may also be determined by using a flag bit. Specifically, for example, as shown in FIG. 5, each service request in the second service request set is represented by a flag bit, that is, one service request corresponds to one flag bit, and a flag bitmap of the second service request set is determined, and the result is obtained. As shown by B in FIG. 5, the black square indicates that the service request corresponding to the flag bit is executed, and the white square indicates that the service request corresponding to the flag bit is not executed; similarly, each service request in the first service request set is also The result is represented by a flag bit, and the result is as shown by C in FIG. 5, and the key of the service request indicated by the same flag bit in B and C is also the same. For example, for the first flag bit in B, the service corresponding to the flag bit is assumed. If the requested key is key1, the key of the service request corresponding to the first flag in C is also key1, and the first flag in B is a white block, indicating that the first flag in C is not executed. A black block indicates that the key1 has been executed. After determining the flag bitmap of the C corresponding to the first service request set by the B corresponding to the second service request set, the service request synchronization apparatus may determine, according to the flag bitmap, the lost service request A that belongs to B but does not belong to C. Ground, can be calculated by the formula: A=BB∩C=B^(B&C), where & indicates that and ^ means XOR, that is, bit and C are first bitwise, and then the result is XORed with B. The flag bit corresponding to the lost service request A can be found. As shown in FIG. 5, the determined lost service request A is the fourth flag bit and the sixth flag bit respectively, and then the reverse search can obtain the specifically lost service. The request determines the service request corresponding to the fourth flag bit and the sixth flag bit.

Alternatively, as an embodiment, the lost service request may also be determined by a bloom filter algorithm. Specifically, the service request in the first service request set and the second service request set is determined. If the service request is a bloom filter, the key is input. For example, if the first service request set includes the input key3, the key3 is Input; calculate the output value corresponding to the input value through a function, for example, by any hash function h(x) Key3 is an input, and the value of the output value h(key3) is calculated; and any number of values of the output value is taken, for example, the last two digits of the output value h(key3) are taken, and the first service request set is obtained according to the two digits. The position belonging to the key 3 is determined in the bitmap, and the bit at the position is set to 1. And so on, each service request in the first service request set and the second service request set is input as input, the position of each service request in the respective bitmap is calculated, and each service is represented in the corresponding bitmap. The request is such that the bitmap of the first service request set and the bitmap of the second service request set are respectively determined, and the two bitmaps are compared, and A is used to indicate that the first storage node needs to be lost. B indicates the service request of the second service request set, and C indicates the service request of the first service request set, that is, the lost service request A belonging to B but not belonging to C can pass the formula: A=BB∩C=B^ (B&C) to calculate, where & indicates that and ^ means XOR, that is, bit and C are first bitwise, and then the result is XORed with B, so that the position of the bitmap corresponding to the lost service request A can be found. . Then, through the inverse calculation of the function, the service request corresponding to the location is obtained as a lost service request; or when the bitmap is generated, and the service request corresponding to each bit of 1 is simultaneously recorded, the relationship of the record can be adopted. The lost business request A is directly determined.

In S130, the service request synchronization device instructs the second storage node to send the lost service request to the first storage node. Specifically, the service request synchronization device may determine, according to the first service request set and the second service request set, which service requests belong to the first storage node in the second log of the second storage node during the failure of the first storage node. The service request executed, these service requests are the service requests lost by the first storage node. The service request synchronization device may instruct the second storage node to send the lost service request to the first storage node, and then execute the lost service request by the first storage node, thereby synchronizing the service request, and improving the storage system to perform the service request. Throughput.

In the embodiment of the present invention, the second storage node may send, according to the indication of the service request synchronization device, the second storage node to directly send the lost service request belonging to the first storage node recorded in the second log to the first storage node. Or the second storage node sends the lost service request to the service request synchronization device, and the service request synchronization device sends the lost service request to the first storage node. Alternatively, after the fault is recovered, the first storage node may send request information to the service request synchronization device, where the request information is used to request a lost service request, and the service request synchronization device instructs the second storage node to the first storage according to the request information. The node sends a lost service request. Or the service request synchronization device instructs the second storage node to send the lost service request to the first storage node, where the service request synchronization device instructs the first storage node to send to the second storage node Sending request information, the request information is used to request a lost service request, and the second storage node sends the lost service request to the first storage node according to the request information. Or the first storage node sends request information to the second storage node, where the request information is used to request a lost service request, and the service request synchronization device instructs the second storage node to send the lost information to the first storage node according to the request information. The business request including the embodiment of the present invention is not limited thereto.

It should be understood that, in various embodiments of the present invention, the size of the sequence numbers of the above processes does not mean the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not be directed to the embodiments of the present invention. The implementation process constitutes any limitation.

Therefore, in the method for storing a node synchronization service request in the embodiment of the present invention, two storage nodes in the storage system may not perform the service request in the order in which the service request is delivered, and may fail when one of the storage nodes fails. Finding service requests that are not recorded by the failed storage node record in different time periods but the service requests belonging to the normal storage node record are lost service requests, so that different storage nodes can execute service requests concurrently or out of order without specialization The allocation and maintenance of the serial number of the service request, and also can accurately obtain the lost service request from the normal storage node, complete the synchronization of the service request, and improve the throughput of the storage system to perform the service request.

A method for storing a node synchronization service request according to an embodiment of the present invention is described in detail above with reference to FIG. 1 to FIG. 5, and a storage node synchronization service according to an embodiment of the present invention will be described below with reference to FIG. 6 to FIG. Requested device.

6 shows a schematic block diagram of an apparatus 200 for storing a node synchronization service request, which may be located in a storage system including a first storage node and a second storage node, the first storage, in accordance with an embodiment of the present invention. The node is configured to execute a plurality of service requests in a first order and store a first log, where the first log is used to record each service request that has been executed in the multiple service requests and start execution of each executed business request At a time, the second storage node is configured to execute the plurality of service requests in a second order and store a second log, the second log is used to record each service request that has been executed in the plurality of service requests and each executed At the start of execution of the business request, the apparatus 200 includes:

The first determining unit 210 is configured to determine, when the first storage node fails within t ₁ to t ₂ of executing the multiple service requests, the first service request set and the second service request set, the first service request The set is a set of service requests recorded in the first log from t ₁ -2dt to t ₂ +2dt, and the second set of service requests is the start execution time recorded in the second log at t ₁ - a set of service requests in dt to t ₂ +dt, the dt being greater than or equal to a difference between a time at which the first storage node begins to execute the target service request and a time at which the second storage node begins to execute the target service request, The target service request is a service request in the first service request set or the second service request set;

The second determining unit 220 is configured to determine, by the service request that belongs to the second service request set that does not belong to the first service request set, a service request that is lost by the first storage node during a failure;

The processing unit 230 is configured to instruct the second storage node to send the lost service request to the first storage node.

Therefore, in the embodiment of the present invention, the device for storing the node synchronization service request may perform the service request in the order in which the two storage nodes in the storage system may not be delivered in the order in which the service request is delivered. By searching for service requests that are not recorded by the failed storage node in different time periods but belonging to the normal storage node record, the service request is a lost service request, so that different storage nodes can execute the service request concurrently or out of order without The serial number of the service request is required to be allocated and maintained, and the lost service request can be accurately obtained from the normal storage node, the service request synchronization is completed, and the throughput of the storage system executing the service request is improved.

Optionally, the first determining unit 210 is specifically configured to: separately sort the first service request set and the service request in the second service request set according to a key of the service request; and compare the first result according to the sorted result. The service request set and the service request in the second service request set obtain a service request that belongs to the second service request set and does not belong to the first service request set, and is the lost service request.

Optionally, the first determining unit 210 is configured to: determine, according to a key of each service request in the first service request set, a flag bitmap of the first service request set; the service request synchronization device is configured according to the second a key of each service request in the service request set, determining a flag bitmap of the second service request set; and obtaining a flag bitmap corresponding to the first service request set and the flag bitmap corresponding to the second service request set, The service request belonging to the second service request set and not belonging to the first service request set is the lost service request.

Optionally, the first determining unit 210 is specifically configured to: determine, according to a Bloom filter bloom filter algorithm and a key of each service request in the first service request set, a bitmap of the first service request set; The service request synchronization device determines a bitmap of the second service request set according to the bloom filter algorithm and a key of each service request in the second service request set; the service request is synchronously loaded Comparing the bitmap of the first service request set with the bitmap of the second service request set, determining that the service request belonging to the second service request set and not belonging to the first service request set is the lost service request.

Optionally, the first sequence may be different from the second sequence or may be the same.

It should be understood that the apparatus 200 for storing a node synchronization service request according to an embodiment of the present invention may correspond to the above method of performing the method 100 in the embodiment of the present invention, and for storing the node synchronization service request. Other operations and/or functions are respectively implemented in order to implement the corresponding processes of the respective methods in FIG. 2, and are not described herein again for brevity.

As shown in FIG. 7, an embodiment of the present invention further provides an apparatus 300 for storing a node synchronization service request, including a processor 310, a memory 320, and a bus system 330. The apparatus 300 is located at a first storage node and a second. In the storage system of the storage node, the first storage node is configured to execute a plurality of service requests in a first order and store a first log, where the first log is used to record each service request that has been executed in the multiple service requests and At the start execution time of each executed business request, the second storage node is configured to execute the plurality of service requests in a second order and store a second log, where the second log is used to record that the plurality of service requests have been recorded Each business request executed and the start execution time of each business request that has been executed.

The processor 310 and the memory 320 are connected by a bus system 330 for storing instructions, and storing a first log and a second log, and the processor 310 is configured to execute the instructions stored by the memory 320. The memory 320 stores program code, and the processor 310 can call the program code stored in the memory 320 to perform the following operations: when the first storage node fails within t ₁ to t ₂ of executing the plurality of service requests, determining a service request set and a second service request set, the first service request set being a set of service requests recorded in the first log starting from t ₁ -2dt to t ₂ +2dt, the second service request And a set of service requests recorded in the second log at a start execution time in t ₁ -dt to t ₂ +dt, where the dt is greater than or equal to a time at which the first storage node starts to execute the target service request and the second a difference between the time when the storage node starts to execute the target service request, the target service request is a service request in the first service request set or the second service request set; determining that the second service request set belongs to the second service request set The service request of the first service request set is a service request lost by the first storage node during a failure; indicating the second storage node to the first storage node Send the missing service request.

It should be understood that, in the embodiment of the present invention, the processor 310 may be a central processing unit ("CPU"), and the processor 310 may also be other general-purpose processors, digital signal processors (DSPs). , an application specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, and the like. The general purpose processor may be a microprocessor or the processor or any conventional processor or the like.

The memory 320 can include read only memory and random access memory and provides instructions and data to the processor 310. A portion of the memory 320 may also include a non-volatile random access memory. For example, the memory 320 can also store information of the device type.

The bus system 330 may include a power bus, a control bus, a status signal bus, and the like in addition to the data bus. However, for clarity of description, various buses are labeled as bus system 330 in the figure.

In the implementation process, each step of the foregoing method may be completed by an integrated logic circuit of hardware in the processor 310 or an instruction in a form of software. The steps of the method disclosed in the embodiments of the present invention may be directly implemented as a hardware processor, or may be performed by a combination of hardware and software modules in the processor. The software module can be located in a conventional storage medium such as random access memory, flash memory, read only memory, programmable read only memory or electrically erasable programmable memory, registers, and the like. The storage medium is located in the memory 320, and the processor 310 reads the information in the memory 320 and combines it The hardware completes the steps of the above method. To avoid repetition, it will not be described in detail here.

Optionally, as an embodiment, the processor 310 may invoke the program code stored in the memory 320 to perform the following operations: respectively, the first service request set and the service request in the second service request set are performed according to the key of the service request. Sorting; according to the sorted result, the service request in the first service request set and the second service request set is obtained, and the service request belonging to the second service request set and not belonging to the first service request set is the lost Business request.

Optionally, as an embodiment, the processor 310 may call the program code stored in the memory 320 to perform the following operations: determining, according to a key of each service request in the first service request set, a flag bit of the first service request set. Determining, according to the key of each service request in the second service request set, a flag bitmap of the second service request set; corresponding to the flag bit corresponding to the first service request set and the second service request set The flag bit obtains a service request belonging to the second service request set and does not belong to the first service request set as the lost service request.

Optionally, as an embodiment, the processor 310 may invoke the program code stored in the memory 320 to perform the following operations: determining, according to the Bloom filter blow filter algorithm and the key of each service request in the first service request set. a bitmap of the first service request set; determining a bitmap of the second service request set according to the bloom filter algorithm and a key of each service request in the second service request set; comparing bits of the first service request set And a bitmap of the second service request set, determining that the service request belonging to the second service request set and not belonging to the first service request set is the lost service request.

It should be understood that the apparatus 300 for storing a node synchronization service request according to an embodiment of the present invention may correspond to the apparatus 200 for storing a node synchronization service request in the embodiment of the present invention, and may correspond to performing according to an embodiment of the present invention. The above-described and other operations and/or functions of respective modules in the method 100 for storing the node synchronization service request, respectively, in order to implement the respective processes of the various methods in FIG. 2, for the sake of brevity, Let me repeat.

Therefore, in the embodiment of the present invention, the device for storing the node synchronization service request may perform the service request in the order in which the two storage nodes in the storage system may not be delivered in the order in which the service request is delivered. By searching for service requests that are not recorded by the failed storage node in different time periods but belonging to the normal storage node record, the service request is a lost service request, so that different storage nodes can execute the service request concurrently or out of order without Need to specifically allocate and maintain the serial number of the service request, and also be able to accurately get from the normal storage node The loss of the business request, the completion of the business request synchronization, improve the throughput of the storage system to perform business requests.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods for implementing the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the present invention.

A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the system, the device and the unit described above can refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.

In the several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

The functions may be stored in a computer readable storage medium if implemented in the form of a software functional unit and sold or used as a standalone product. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, or a part of the technical solution, may be embodied in the form of a software product, which is stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), A variety of media that can store program code, such as random access memory (RAM), disk, or optical disk.

The above is only a specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention. It should be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

A method for storing a node synchronization service request, wherein the method is applied to a storage system including a first storage node and a second storage node, where the first storage node is configured to execute in a first order The service requests and stores a first log, where the first log is used to record each service request that the first storage node has performed in the multiple service requests and each service that the first storage node has performed At the start of execution of the request, the second storage node is configured to execute the plurality of service requests in a second order and store a second log, where the second log is used to record the second of the plurality of service requests Each of the service requests that have been executed by the storage node and the start execution time of each service request that has been executed by the second storage node, the method includes:

When the first storage node fails within t 1 to t 2 of executing the plurality of service requests, the service request synchronization device determines a first service request set and a second service request set, the first service request set a set of service requests for the start execution time in the first log from t 1 -2dt to t 2 +2dt, the second service request set being the start execution time recorded in the second log at t 1 -dt to t 2 + dt request traffic within a set, the dt is greater than or equal to the first storage node begins execution time of the target service request time with the second target storage node starts executing the service request a difference between the target service request is a service request in the first service request set or the second service request set;

The service request synchronization device determines that the service request that belongs to the second service request set and does not belong to the first service request set is a service request that is lost by the first storage node during a failure;

The service request synchronization device instructs the second storage node to send the lost service request to the first storage node.
The method according to claim 1, wherein the service request synchronization means determines that a service request belonging to the second service request set and not belonging to the first service request set is that the first storage node is occurring Business requests lost during a failure, including:

The service request synchronization device sequentially sorts the service request in the first service request set and the second service request set according to a key of the service request;

And the service request synchronization device compares the service request in the first service request set and the second service request set according to the sorted result, and obtains the second service request set that does not belong to the first service request. The aggregated business request is for the lost business request.
The method of claim 1 wherein said service request synchronization device Determining, by the service request that belongs to the second service request set that does not belong to the first service request set, a service request that is lost during the failure of the first storage node, including:

Determining, by the service request synchronization device, a flag bitmap of the first service request set according to a key of each service request in the first service request set;

Determining, by the service request synchronization device, a flag bitmap of the second service request set according to a key of each service request in the second service request set;

The service request synchronization device compares the flag bitmap of the first service request set with the flag bitmap of the second service request set to obtain the second service request set and does not belong to the first service request. The aggregated business request is for the lost business request.
The method according to claim 1, wherein the service request synchronization means determines that a service request belonging to the second service request set and not belonging to the first service request set is that the first storage node is occurring Business requests lost during a failure, including:

The service request synchronization device determines a bitmap of the first service request set according to a Bloom filter and a key of each service request in the first service request set;

Determining, by the service request synchronization device, a bitmap of the second service request set according to the bloom filter algorithm and a key of each service request in the second service request set;

The service request synchronization device compares the bitmap of the first service request set with the bitmap of the second service request set, and determines a service that belongs to the second service request set and does not belong to the first service request set. Request for the lost business request.
The method according to any one of claims 1 to 4, wherein the first order is different from the second order.
An apparatus for storing a node synchronization service request, wherein the apparatus is located in a storage system including a first storage node and a second storage node, where the first storage node is configured to execute multiples in a first order The service requests and stores a first log, where the first day is used to record each service request that the first storage node has performed in the multiple service requests and each service that the first storage node has performed At the start of execution of the request, the second storage node is configured to execute the plurality of service requests in a second order and store a second physique, and the second log is configured to record the second of the plurality of service requests The storage execution time of each service request that has been executed by the node and the start of execution of each service request that the second storage node has performed, the device includes:

A first determining unit, when the first storage node when a failure occurs within 1 to t 2 performs the service request plurality of t, and determining a second set of a first service request service request set, said first The service request set is a set of service requests recorded in the first log at a start execution time from t 1 -2dt to t 2 +2dt, and the second service request set is a start execution of the record in the second log a set of service requests at time t 1 -dt to t 2 +dt, the dt being greater than or equal to a time when the first storage node starts to execute a target service request, and the second storage node starts executing the target service a difference between the requested moments, where the target service request is a service request in the first service request set or the second service request set;

a second determining unit, configured to determine, by the service request that belongs to the second service request set that does not belong to the first service request set, a service request that is lost by the first storage node during a failure;

And a processing unit, configured to instruct the second storage node to send the lost service request to the first storage node.
The device according to claim 6, wherein the first determining unit is specifically configured to:

And respectively sorting the service request in the first service request set and the second service request set according to a key of a service request;

Obtaining, according to the sorted result, the service request in the first service request set and the second service request set, obtaining a service request that belongs to the second service request set and does not belong to the first service request set. Describe the lost business request.
The device according to claim 6, wherein the first determining unit is specifically configured to:

Determining, according to a key of each service request in the first service request set, a flag bitmap of the first service request set;

Determining, according to a key of each service request in the second service request set, a flag bitmap of the second service request set;

Comparing the flag bit corresponding to the first service request set with the flag bit corresponding to the second service request set, obtaining a service request that belongs to the second service request set and does not belong to the first service request set Describe the lost business request.
The device according to claim 6, wherein the first determining unit is specifically configured to:

Determining a bitmap of the first service request set according to a Bloom filter bloom filter algorithm and a key of each service request in the first service request set;

Determining, according to the bloom filter algorithm and a key of each service request in the second service request set, a bitmap of the second service request set;

Comparing the bitmap of the first service request set with the bitmap of the second service request set, determining that the service request belonging to the second service request set and not belonging to the first service request set is the lost Business request.
The apparatus according to any one of claims 6 to 9, wherein the first order is different from the second order.
An apparatus for storing a node synchronization service request, wherein the apparatus is located in a storage system including a first storage node and a second storage node, where the first storage node is configured to execute multiples in a first order The service requests and stores a first log, where the first log is used to record each service request that the first storage node has performed in the multiple service requests and each service request that has been executed by the first storage node The second storage node is configured to execute the plurality of service requests in a second order and store a second log, where the second log is used to record the second storage of the plurality of service requests Each of the service requests that the node has executed and the start execution time of each service request that the second storage node has performed, the device includes: a memory and a processor,

The memory is configured to store the first log and the second log,

The processor is used to:

Determining, when the first storage node fails within t 1 to t 2 of executing the plurality of service requests, the first service request set and the second service request set, where the first service request set is the a set of service requests recorded in a log from t 1 -2dt to t 2 +2dt, the second service request set being the start execution time recorded in the second log at t 1 -dt to a set of service requests within t 2 +dt, the dt being greater than or equal to a difference between a time at which the first storage node begins to execute a target service request and a time at which the second storage node begins to execute the target service request a value, the target service request is a service request in the first service request set or the second service request set;

Determining, by the service request that belongs to the second service request set and not belonging to the first service request set, a service request that is lost during the failure of the first storage node;

Instructing the second storage node to send the lost service request to the first storage node.
The device according to claim 11, wherein the processor is specifically configured to:

Separating the first service request set and the service request in the second service request set respectively Sort by the key of the business request;

Obtaining, according to the sorted result, the service request in the first service request set and the second service request set, obtaining a service request that belongs to the second service request set and does not belong to the first service request set. Describe the lost business request.
The device according to claim 11, wherein the processor is specifically configured to:

Determining, according to a key of each service request in the first service request set, a flag bitmap of the first service request set;

Determining, according to a key of each service request in the second service request set, a flag bitmap of the second service request set;

Comparing the flag bitmap corresponding to the first service request set with the flag bitmap corresponding to the second service request set, obtaining a service request that belongs to the second service request set and does not belong to the first service request set For the lost business request.
The device according to claim 11, wherein the processor is specifically configured to:

Determining a bitmap of the first service request set according to a Bloom filter bloom filter algorithm and a key of each service request in the first service request set;

Determining, according to the bloom filter algorithm and a key of each service request in the second service request set, a bitmap of the second service request set;

Comparing the bitmap of the first service request set with the bitmap of the second service request set, determining that the service request belonging to the second service request set and not belonging to the first service request set is the lost Business request.
Apparatus according to any one of claims 11 to 14, wherein said first order is different from said second order.