WO2013106993A1

WO2013106993A1 - Capacity expansion method and device and data access method and device

Info

Publication number: WO2013106993A1
Application number: PCT/CN2012/070464
Authority: WO
Inventors: 印雨; 潘孝忠
Original assignee: 华为技术有限公司
Priority date: 2012-01-17
Filing date: 2012-01-17
Publication date: 2013-07-25
Also published as: CN103348653B; CN103348653A

Abstract

Embodiments of the present invention relate to the technical field of data storage, and provide a capacity expansion method and device and a data access method and device. The method comprises: determining, according to the number of physical nodes before capacity expansion, the number of virtual nodes, so that each physical node before capacity expansion corresponds to n virtual nodes, n being smaller than a preset value; when physical nodes are added during capacity expansion of the system and a part of the physical nodes each correspond to one virtual node, the virtual nodes before capacity expansion being split to generate new virtual nodes, so that the client re-determines, according to the number of the virtual nodes after splitting, the virtual nodes for data mapping; migrating a part of the virtual nodes to the newly added physical nodes; and determining data managed by the virtual nodes and the newly obtained virtual nodes after splitting, and respectively deleting data not managed by the virtual nodes and the newly obtained virtual nodes after splitting. By means of the above solution, the present invention improves the reliability and the expandability of a storage device.

Description

Method and device for expanding capacity, and method and device for accessing data

The present invention relates to the field of data storage, and in particular, to a method and device for expanding a capacity, and a method and device for accessing data.

Say

Background technique

P2P (Pear to Pear) storage devices based on DHT (Distributed Hash Table) technology have high scalability, support large-scale data storage, and can better cope with human data in the era of network information explosion. Ultra-large storage and management challenges. However, with the growth of storage data, the capacity of storage devices needs to be expanded to meet the growing demand for data storage.

When the storage device is initialized, the storage device is allocated a specified number of virtual nodes in advance, and a mapping relationship between the physical node and the virtual node is established, so that there are multiple virtual nodes on one physical node. When the storage device is expanded, that is, when the physical node increases, the virtual node is dynamically migrated to the new physical node. For example, before the capacity expansion, each physical node assumes 2N virtual nodes. After the capacity expansion, the physical nodes are doubled. If the load is balanced, each physical node generally assumes N virtual nodes. During the system expansion process, the total number of virtual nodes is always constant.

In the process of implementing the present invention, the inventors have found that the prior art has at least the following problems:

In the existing expansion plan, the total number of virtual nodes is preset and remains constant during the system expansion process. In order to ensure the scalability of the system, a large number of virtual nodes are usually set in advance. Before the system is expanded, the number of virtual nodes corresponding to each physical node is large. When the number of data copies is constant, the physical nodes correspond. The more virtual nodes, the worse the reliability of the system; and, although the total number of pre-set virtual nodes is relatively large, as the system continues to expand, it will eventually expand to the point where no virtual nodes can be assigned to new physical nodes, making the system unable to Capacity expansion, system scalability is poor. Summary of the invention

In order to improve the reliability and scalability of the storage device, an embodiment of the present invention provides a method and a storage device for capacity expansion. The technical solution is as follows:

A method of expanding a volume, the method comprising:

Determine the number of virtual nodes based on the number of physical nodes before the expansion, and make the physical nodes before each expansion corresponding to n. Virtual nodes, n is less than the preset value;

When a physical node is added to the system and a physical node corresponds to a virtual node, the virtual node before the expansion is split, and a new virtual node is generated, so that the client re-determines according to the number of the virtual nodes after the split. Virtual node of data mapping;

Migrate some virtual nodes to newly added physical nodes;

The data managed by the virtual node and the new virtual node obtained by the splitting is determined, and the data that is not managed by the virtual node and the new virtual node that is split is deleted.

A storage device, the storage device includes:

a determining module, configured to determine the number of virtual nodes according to the number of physical nodes before the expansion, so that the physical nodes before each expansion correspond to n virtual nodes, where n is less than a preset value;

The splitting module is used to expand the physical node when the system is expanded, and some physical nodes appear corresponding to one virtual node. The virtual node before the expansion is split, and a new virtual node is generated, so that the client is based on the split virtual node. Quantity, redefining the virtual node of the data map;

a migration module, configured to migrate part of the virtual node to the newly added physical node;

The deletion module is configured to determine data managed by the virtual node and the new virtual node that is split by the virtual node, and delete the data that is not managed by the virtual node and the new virtual node that is split by the virtual node. Based on the foregoing expansion method, a method for accessing data is also provided, where the method includes:

When the virtual node is split, the corresponding relationship between the physical node and the virtual node is updated, and the first correspondence is obtained. The virtual node that determines the data mapping to be accessed is determined according to the first correspondence, and the first physical node where the data is located is determined. Requiring access data to the first physical node;

Updating the correspondence between the physical node and the virtual node to obtain a second correspondence relationship when the virtual node is migrated; determining, according to the second correspondence, the second physical node where the data is located, and requesting access to the second physical node data.

A client device, the client device includes:

An update module, configured to update a correspondence between a physical node and a virtual node when the virtual node is split, to obtain a first correspondence relationship;

An access module, configured to determine a virtual node of the data mapping to be accessed, determining, according to the first correspondence, a first physical node where the data is located, and requesting access data from the first physical node;

The update module is further configured to: when the virtual node migrates, update a correspondence between the physical node and the virtual node to obtain a second correspondence; The access module is further configured to determine, according to the second correspondence, a second physical node where the data is located, and request access data from the second physical node.

The beneficial effects of the technical solutions provided by the embodiments of the present invention are:

In the case that each physical node corresponds to a small number of virtual nodes, the virtual node is dynamically migrated by the virtual node splitting during the capacity expansion, which avoids the reliability problem caused by pre-setting a large number of virtual nodes, and improves the storage device. The reliability, while avoiding the problem of non-expansion caused by the fact that no virtual nodes can be allocated, improves the scalability of the storage device. And, based on the foregoing expansion method, a corresponding access data scheme is provided. DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described. It is obvious that the drawings in the following description are only some embodiments of the present invention. Other drawings may also be obtained from those of ordinary skill in the art in view of the drawings.

1 is a flowchart of a method for expanding capacity provided by Embodiment 1 of the present invention;

2 is a flowchart of a method for expanding capacity provided by Embodiment 2 of the present invention;

3 is a schematic diagram of a partition of a pre-split memory device according to Embodiment 2 of the present invention;

4 is a schematic diagram of a partition of a storage device after splitting according to Embodiment 2 of the present invention;

5 is a flowchart of a method for accessing data provided by Embodiment 3 of the present invention;

6 is a schematic diagram of a process of writing data by a client device according to Embodiment 3 of the present invention;

7 is a schematic diagram of a process of reading data by a client device according to Embodiment 3 of the present invention;

8 is a schematic structural diagram of a storage device according to Embodiment 4 of the present invention;

FIG. 9 is a schematic structural diagram of a client device according to Embodiment 5 of the present invention. detailed description

The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

Example 1

Referring to FIG. 1, this embodiment provides a method for expanding a capacity, which may be performed by a storage device, where the method includes:

101: Determine, according to the number of physical nodes before the expansion, the number of virtual nodes, so that the physical nodes before each expansion correspond to n virtual nodes, where n is smaller than a preset value;

102: When a system is expanded, a physical node is newly added, and a part of the physical node corresponds to a virtual node. When the virtual node is expanded before the expansion, a new virtual node is generated, and the client re-determines the virtual node of the data mapping according to the number of the virtual nodes after the splitting;

103: Migrate part of the virtual node to the newly added physical node;

104: Determine data respectively managed by the virtual node and the new virtual node that is split by the virtual node, and delete the data that is not managed by the virtual node and the new virtual node that is split by the virtual node.

In this embodiment, when each physical node corresponds to a small number of virtual nodes, the virtual node is dynamically migrated by the virtual node splitting during the capacity expansion, thereby avoiding the reliability problem caused by presetting a large number of virtual nodes, and improving The reliability of the storage device is avoided, and the problem of non-expansion caused by the fact that no virtual node can be allocated is avoided, and the scalability of the storage device is improved. Example 2

Referring to FIG. 2, this embodiment provides a method for expanding a capacity, which may be performed by a storage device. The method includes: 201: Determine the number of virtual nodes according to the number of physical nodes before the expansion, so that each expansion is performed. The physical node corresponds to n virtual nodes, and n is smaller than a preset value;

The default value is the number of virtual nodes corresponding to each physical node in the prior art. The typical value of n in this embodiment may be 2 or 3. The typical value is usually much smaller than the number of virtual nodes corresponding to each physical node in the prior art. Referring to Figure 3, taking 5 physical nodes as an example, there are 10 virtual nodes, which are P0, Pl, P2, P3, P4, P5, P6, P7, P8, and P9, and each physical node corresponds to 2 virtual nodes. The correspondence between the physical node and the virtual node is as follows:

Physical node 1 : P0, P5

Physical node 2: Pl, P6

Physical node 3: P2, P7

Physical node 4: P3, P8

Physical node 5 : P4, P9

When each data has M data backups (or copies), when a physical node fails, the concurrency can reach n X (M-1 ), that is, the n virtual nodes on the failed physical node can concurrently ( Data is recovered on M-1) backup nodes. By reducing the number of virtual nodes corresponding to each physical node, the reliability of the system can be improved. This is because the reliability of the system has a serious dependence on the number of virtual nodes corresponding to each physical node. If a physical node fails, the physical node of the replica of all the virtual nodes corresponding to the failed physical node fails at the same time. Will cause the system to be unavailable. Therefore, reducing the number of virtual nodes corresponding to each physical node and reducing the degree of concurrency can improve the reliability of the system. 202: When a physical node is newly added when the system is expanded, and a part of the physical node corresponds to one virtual node, the virtual node before the expansion is split, and a new virtual node is generated;

Referring to the split diagram shown in Figure 4, let the virtual node be Pi, before the split i = 0, 1, 2, ..., 9, the total number of virtual nodes before splitting N = 10, each virtual node splits a new virtual node Pi splits P (i+N), that is, P0 splits P10, PI splits Pl l, and so on, P9 splits P19, and the total number of virtual nodes after splitting is 2N=20. After the splitting, the virtual node corresponding to each physical node is doubled. Take Figure 3 as an example. After the split, each physical node corresponds to four virtual nodes. The corresponding relationship between the physical node and the virtual node after splitting is as follows:

Physical node 1 : P0, P5, P10, P15

Physical node 2: Pl, P6, Pl l, P16

Physical node 3: P2, P7, P12, P17

Physical node 4: P3, P8, P13, P18

Physical node 5: P4, P9, P14, P19

It should be noted that each virtual node and its split new virtual node point to the same physical node data, that is, Pi and P (i+N) jointly point to the data on the same physical node, for example, P0 and P10 points to the same physical node, and P1 and P11 point to the same physical node.

Since the virtual nodes corresponding to each physical node before the splitting are evenly distributed, the virtual nodes corresponding to each physical node are still balanced after the splitting. Therefore, the read and write requests to the client do not cause data hotspot problems.

203: Re-determine the virtual node of the data mapping according to the number of the virtual nodes after the splitting;

Specifically, the hash of the data is hashed to obtain a hash value, and the hash value is modulo the total number of the split virtual nodes to obtain a modulus value, and the virtual node corresponding to the data is determined according to the modulus value. The formula is expressed as follows:

Hash (key) mod (2N) ,

Hash represents a hash operation, key represents the identity of the data, mod represents a modulo operation, and N is the number of virtual nodes before the split. Correspondingly, 2N represents the number of virtual nodes after the split.

Since the virtual node and the new virtual node obtained by the split point to the same data, the above remapping process does not cause data pointing errors. For example, Hash (abc) mod (N) =0, that is, the data before the split key abc is mapped to the virtual node P0, Hash (abc) mod (2N) = 10, that is, the data after the split key abc is mapped to The virtual node P10, since the virtual node P10 is split by the virtual node P0, the virtual node P10 and the virtual node P0 point to the same data, and therefore, the data pointing error is not caused in the remapping process.

204: Migrate part of the virtual node to the newly added physical node;

In this embodiment, the migration policy is not specifically limited, and any strategy for enabling the virtual nodes corresponding to the physical nodes to be substantially averaged is adapted to the embodiment. For example, calculate the average of the number of virtual nodes corresponding to each physical node, and the physical section The virtual node corresponding to the virtual node number exceeding the average value is migrated to the physical node whose corresponding virtual node number is smaller than the average value. The following description will be made in conjunction with a specific example.

After splitting, each physical node corresponds to an average of 20/5=4 virtual nodes. When the system is expanded, two new physical nodes are added, and a total of seven physical nodes are added. After expansion, each physical node corresponds to an average of 20/7 «2. 85 logics. Nodes, that is, some physical nodes correspond to three virtual nodes, and some physical nodes correspond to two virtual nodes. To reduce the amount of data migration, a total of five virtual nodes are migrated to the newly added physical node, and one physical node and one virtual node are migrated. The corresponding relationship can be: Physical node 1: P0, P5, P10

Physical node 2: Pl, P6, P16

Physical node 3: P2, P12, P17

Physical node 4: P8, P13, P18

Physical node 5: P4, P14, P19

Physical node 6: P15, Pl l, P7

Physical node 7: P3, P9

205: Determine data respectively managed by the virtual node and the new virtual node that is split by the virtual node, and delete data that is not managed by the virtual node and the new virtual node that is split by the virtual node.

Since the virtual node and the new virtual node obtained by its split point to the same data, half of the data on the virtual node and the new virtual node that is split is invalid. Therefore, the virtual node and its split are deleted after the migration. The obtained new virtual node does not belong to its own management data, which can improve the utilization of the disk.

In this embodiment, when each physical node corresponds to a small number of virtual nodes, the virtual node is dynamically migrated by the virtual node splitting during the capacity expansion, thereby avoiding the reliability problem caused by presetting a large number of virtual nodes, and improving The reliability of the storage device is avoided, and the problem of non-expansion caused by the fact that no virtual node can be allocated is avoided, and the scalability of the storage device is improved. Example 3

The method for accessing data according to the embodiment of the present invention is as follows: Referring to FIG. 5, the embodiment provides a method for accessing data, and the method may be performed by a client device, where the method includes:

301: When the virtual node is split, update the correspondence between the physical node and the virtual node to obtain a first correspondence.

302: Determine a virtual node of the data mapping to be accessed, determine, according to the first correspondence, a first physical node where the data is located, and request access data from the first physical node;

303: When the virtual node migrates, update the correspondence between the physical node and the virtual node to obtain a second correspondence.

304: Determine, according to the second correspondence, the second physical node where the data is located, and request the second physical node to visit Ask the data.

After requesting access to the second physical node at step 304, if the data has not been migrated from the first physical node to the second physical node, the second physical node acquires the data from the first physical node.

Before step 301, the method further includes:

Determining a virtual node of the data mapping to be accessed, determining, according to the third correspondence, the first physical node where the data is located, requesting access data from the first physical node, where the third correspondence relationship is before the virtual node is split, the physical node and the virtual node The correspondence of nodes.

The following takes the client device write data process (put) and the read data process (get) as an example.

See the client device write data process shown in Figure 6, which includes pre-split, split, and the entire data write process of migrating the virtual node to the newly added physical node. For example, the correspondence between physical nodes and virtual nodes in each phase is as follows:

Before splitting (third correspondence) After splitting (first-corresponding relationship) After migration (second correspondence) Physical node 1: P0, P5 Physical node 1: P0, P5, P10, P15 Physical node 1: P0, P5 P10 Physical node 2: Pl, P6 Physical node 2: Pl, P6, Pl l, P16 Physical node 2: Pl, P6, P16 Physical node 3: P2, P7 Physical node 3: P2, P7, P12, P17 Physical node 3: P2, P12, P17 Physical Node 4: P3, P8 Physical Node 4: P3, P8, P13, P18 Physical Node 4: P8, P13, P18 Physical Node 5: P4, P9 Physical Node 5: P4, P9, P14 P19 Physical node 5: P4, P14, P19 Physical node 6: P15, Pl l, P7 Physical node 7: P3, P9 401: The client device determines the virtual node of the data mapping to be written, according to the third correspondence, Determining the physical node where the data is located, requesting write data to the physical node;

The third correspondence relationship is a correspondence between the physical node and the virtual node before the virtual node is split.

Set the data to be written key=abc, value=lll. Determine the virtual node of the data map according to the formula Hash (key) mod (N). Let Hash (abc) =15, Hash (abc) mod (10) =5, that is, the data is mapped to the virtual node P5, according to the third correspondence, the data requested to be written (key= _a bc, value=lll) The physical node is physical node 1.

402: The virtual node is split, and the client device updates the correspondence between the physical node and the virtual node to obtain a first correspondence.

For the virtual node splitting process, refer to step 202, and details are not described herein again.

403: The client device re-determines the virtual node of the data mapping to be written, determines a physical node where the data is located according to the first correspondence, and requests the physical node to write data.

According to the formula Hash (key) mod (2N), the physical node where the data is located after the split is determined. Let Hash (abc) = 15, Hash (abc) mod (20) = 15, that is, the data is mapped to the virtual node P15. According to the first correspondence, the physical node where the data requested to be written (key=abc, value=lll) is still the physical node 1 . At the same time, since P15 is obtained by splitting P5, the data of the client device is written to the correct virtual node on the correct physical node.

404: The virtual node is migrated, and the client device updates the correspondence between the physical node and the virtual node to obtain a second correspondence.

For the virtual node migration process, refer to step 204, and details are not described herein. For example, virtual node P15 is migrated from physical node 1 to physical node 6.

405: The client device determines, according to the second correspondence, the physical node where the data is located, and requests the physical node to write data.

The data requested to be written (key=abc, value=lll) is mapped to the virtual node P15. According to the correspondence between the physical node and the virtual node after the migration, the virtual node P15 requests the physical node 6 to write data to the physical node 6. . See the client device read data process shown in Figure 7, which includes pre-splitting, splitting, and the entire data reading process of migrating the virtual node to the newly added physical node. It is assumed that the correspondence between the physical node and the virtual node in each stage of the process of reading data by the client device is the same as that in the process of writing data by the client device.

501: The client device determines a virtual node of the data mapping to be read, determines a physical node where the data is located according to the third correspondence, and requests the physical node to read the data.

Set the data to be read key=abc, value=l l l. Determine the virtual node of the data map according to the formula Hash (key) mod (N). Let Hash (abc) mod (10) = 5, that is, the data is mapped to the virtual node P5. According to the third correspondence, the physical node where the data requested to be read (key=abc, value=lll) is the physical node 1 .

502: The virtual node is split, and the client device updates the correspondence between the physical node and the virtual node to obtain a first correspondence.

503: The client device re-determines the virtual node of the data mapping to be read, determines a physical node where the data is located according to the first correspondence, and requests the physical node to read the data.

According to the formula Hash (key) mod (2N) , the physical node where the data is located after the split is determined. Let Hash (abc) mod (20) = 15, that is, the data is mapped to the virtual node P15. According to the first correspondence, the physical node where the data requested to be read (key=abc, value=lll) is still a physical node. 1. At the same time, since P15 is obtained by splitting P5, the client device can correctly read the data.

504: The virtual node is migrated, and the client device updates the correspondence between the physical node and the virtual node to obtain a second correspondence. For the virtual node migration process, refer to step 204, and details are not described herein. For example, the virtual node P15 is migrated from the physical node 1 to the physical node 6.

505: The client device determines, according to the virtual node mapped by the data to be read, and the second corresponding relationship, the physical node where the data is located, and requests the physical node to read the data.

The data requested to be read (key= _a bc, value=l11) is mapped to the virtual node P15. According to the second correspondence, the virtual node P15 requests the physical node 6 to read data from the physical node 6.

Further, if the data to be read has not been migrated from the previous physical node to the physical node, the physical node redirects (or forwards) the request for reading the data to the previous physical node, and from the previous The physical node obtains the data to ensure that the client device can still read the data.

For example, if the data to be read has not been migrated from the physical node 1 to the physical node 6, the physical node 6 redirects the request to the physical node 1 after receiving the request from the client device to read the data.

This embodiment provides a corresponding access data scheme based on the capacity expansion method provided by the embodiment of the present invention. In addition, in the virtual node splitting process, the virtual node and the new virtual node that is split are the same physical node. Therefore, during the capacity expansion process, the client device can still access the data correctly, and the upper layer service is not affected. Example 4

Referring to FIG. 8, the embodiment provides a storage device, where the storage device includes:

The determining module 601 is configured to determine the number of virtual nodes according to the number of physical nodes before the expansion, so that the physical nodes before each expansion correspond to n virtual nodes, where n is smaller than a preset value;

The splitting module 602 is configured to: when a physical node is newly added when the system is expanded, and a part of the physical node corresponds to a virtual node, the virtual node before the expansion is split, and a new virtual node is generated, so that the client is virtualized according to the split. The number of nodes, re-determining the virtual nodes of the data mapping;

a migration module 603, configured to migrate a part of the virtual node to the newly added physical node;

The deleting module 604 is configured to determine data respectively managed by the virtual node and the new virtual node that is split by the virtual node, and delete data that is not managed by the virtual node and the new virtual node that is split by the virtual node.

Wherein, the migration module 603 is specifically used for

The average value of the number of virtual nodes corresponding to each physical node is calculated, and the virtual node whose physical node corresponds to the average number of virtual nodes is migrated to the physical node whose physical node corresponds to the number of virtual nodes less than the average.

In this embodiment, when each physical node corresponds to a small number of virtual nodes, the virtual node is dynamically migrated by the virtual node splitting during the capacity expansion, thereby avoiding the reliability problem caused by presetting a large number of virtual nodes, and improving The reliability of the storage device, while avoiding the problem of non-expansion caused by the fact that no virtual node can be allocated, improving the storage The scalability of the storage device. Example 5

Referring to FIG. 9, this embodiment provides a client device, where the client device includes:

The updating module 701 is configured to update a correspondence between the physical node and the virtual node when the virtual node is split, to obtain a first correspondence relationship;

The access module 702 is configured to determine a virtual node of the data mapping to be accessed, determine, according to the first correspondence, the first physical node where the data is located, and request the access data from the first physical node;

The update module 701 is further configured to: when the virtual node migrates, update the correspondence between the physical node and the virtual node to obtain a second correspondence;

The access module 702 is further configured to determine, according to the second correspondence, the second physical node where the data is located, and request the access data from the second physical node.

The access module 702 includes: a determining unit, configured to determine a virtual node of the data mapping to be accessed according to the formula Hash (key) mod (2N),

Hash represents a hash algorithm, key represents the identifier of the data to be accessed, mod represents the modulo operation, N is the number of virtual nodes before the split, and 2N represents the number of virtual nodes after the split.

Further, the accessing module 702 is further configured to: before updating the correspondence between the physical node and the virtual node when the virtual node is split, determine the virtual node of the data mapping to be accessed, and determine the first physical location where the data is located according to the third correspondence relationship. The node requests access data from the first physical node, and the third correspondence relationship is a correspondence between the physical node and the virtual node before the virtual node is split.

This embodiment provides a corresponding access data scheme. In addition, in the virtual node splitting process, the virtual node and the new virtual node that is split are the same physical node. Therefore, during the capacity expansion process, the client device can still access the data correctly, and the upper layer service is not affected. The storage device and the client device provided by the present invention are the same as the method embodiment, and those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the device and the unit described above may refer to the foregoing. Corresponding processes in the method embodiments are not described herein again. It will be apparent to those skilled in the art that the device embodiments described above are merely illustrative, and the division of the units/modules is only a logical function division, and may be further divided in actual implementation. For example, each functional unit/module can be integrated in one processing unit/module, or each unit/module can be physically stored separately. It is also possible to integrate two or more units/modules in one unit/module. The above integrated unit/module can be implemented in the form of hardware or in the form of a software functional unit/module. A person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium. The storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like. The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., which are within the spirit and scope of the present invention, should be included in the protection of the present invention. Within the scope.

Claims

Claim

A method for expanding a capacity, the method comprising:

Determine the number of virtual nodes according to the number of physical nodes before the expansion, so that the physical nodes before each expansion correspond to _n virtual nodes, where n is smaller than the preset value;

Migrate some virtual nodes to newly added physical nodes;

The data managed by the virtual node and the new virtual node obtained by the splitting is determined, and the data that is not managed by the virtual node and the new virtual node that is split by the virtual node is deleted.

The method according to claim 1, wherein the migrating part of the virtual node to the newly added physical node includes:

The average value of the number of virtual nodes corresponding to each physical node is calculated, and the virtual node corresponding to the physical node corresponding to the virtual node is migrated to the physical node whose physical node corresponds to the virtual node whose number is smaller than the average value.

A method for accessing data according to the expansion method of claim 1, wherein the method comprises: updating a correspondence between a physical node and a virtual node when the virtual node is split, to obtain a first correspondence; Determining a virtual node of the data mapping to be accessed, determining, according to the first correspondence, a first physical node where the data is located, and requesting access data from the first physical node;

The method according to claim 3, wherein the determining the virtual node of the data mapping to be accessed comprises:

Determine the virtual node of the data map to be accessed according to the formula Hash (key) mod (2N),

The method according to claim 3, wherein the method further includes: before updating the correspondence between the physical node and the virtual node when the virtual node is split, the method further includes: Determining a virtual node of the data mapping to be accessed, determining, according to the third correspondence, the first physical node where the data is located, requesting access data from the first physical node, where the third correspondence is before the virtual node is split. The correspondence between physical nodes and virtual nodes.

The method according to claim 3, wherein after the requesting the second physical node to access the data, the method further includes:

If the data has not been migrated from the first physical node to the second physical node, the second physical node acquires the data from the first physical node.

A storage device, wherein the storage device comprises:

a determining module, configured to determine the number of virtual nodes according to the number of physical nodes before the expansion, so that the physical nodes before each expansion correspond to n virtual nodes, where n is smaller than a preset value;

The deletion module is configured to determine data managed by the virtual node and the new virtual node that is split by the virtual node, and delete the data that is not managed by the virtual node and the new virtual node that is split by the virtual node.

The storage device according to claim 7, wherein the migration module is specifically configured to calculate an average value of the number of virtual nodes corresponding to each physical node, and the number of virtual nodes corresponding to the physical node exceeds the average value. The virtual node is migrated to a physical node whose physical node corresponds to a virtual node whose number is smaller than the average value.

A client device, the client device comprising:

An update module, configured to update a correspondence between a physical node and a virtual node when the virtual node is split, to obtain a first corresponding relationship;

The update module is further configured to update a correspondence between the physical node and the virtual node when the virtual node migrates, to obtain a second correspondence relationship;

The access module is further configured to determine, according to the second correspondence, a second physical node where the data is located, and request access data from the second physical node.

The client device according to claim 9, wherein the access module comprises: a determining unit, configured to:

The client device according to claim 9, wherein the access module is further configured to: before updating the correspondence between the physical node and the virtual node when the virtual node is split,

Determining a virtual node of the data mapping to be accessed, determining, according to the third correspondence, the first physical node where the data is located, requesting access data from the first physical node, where the third correspondence is before the virtual node is split. The correspondence between physical nodes and virtual nodes.