A copy distribution system, method, apparatus, storage medium, and electronic device.

Abstract

This specification discloses a replica distribution system, method, apparatus, storage medium, and electronic device. In a distributed database system, each database server reports its network latency to each landmark node to a load balancing server. The load balancing server selects a target server based on the reported network latency and notifies the database server storing the primary replica corresponding to the target partition to distribute the secondary replicas corresponding to the primary replica to the target server. This method can distribute replicas corresponding to a partition based on the network latency of each database server, enabling flexible deployment of replicas according to the specific network topology of the distributed database system, effectively improving the query efficiency of the distributed database system.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and in particular to a copy distribution system, method, apparatus, storage medium, and electronic device. Background Technology

[0002] With the development of the times, the demand for data storage has increased dramatically. Currently, distributed database systems are often used to meet the storage needs of ultra-large-scale data.

[0003] Generally, in order to improve the query efficiency and reliability of a distributed database system, the data table is usually partitioned. For example, data whose keys start with the letter a to m can be divided into one partition, and data whose keys start with the letter n to z can be divided into another partition. Different copies of the partitions are stored on different database servers in the distributed database system to distribute the query pressure of the same data table.

[0004] Therefore, how to manage these replicas to further improve the query efficiency of distributed database systems and cope with the increasing demand for high-concurrency queries is an urgent problem to be solved. Summary of the Invention

[0005] This specification provides a copy distribution system, method, apparatus, storage medium, and electronic device to partially address the problems existing in the prior art.

[0006] The embodiments in this specification adopt the following technical solutions:

[0007] This specification provides a copy distribution system, which includes a load balancing server and several database servers; wherein:

[0008] Any database server is used to determine the network latency between itself and each preset landmark node, and report the network latency to the load balancing server.

[0009] The load balancing server is used to receive network latency reported by each database server, determine the target server among each database server based on the network latency between each database server and each landmark node, and send the identifier of the target server to the database server that stores the primary replica corresponding to the target partition.

[0010] A database server storing the primary replicas corresponding to the target partition is used to receive the identifier of the target server and distribute the secondary replicas corresponding to the primary replicas to the target server.

[0011] This specification provides a method for distributing copies, the method comprising:

[0012] Receive network latency reports from each database server, where the network latency is the network latency between each database server and each preset landmark node;

[0013] Based on the network latency, the target server is determined among the database servers;

[0014] The identifier of the target server is sent to the database server that stores the primary replica corresponding to the target partition, so that the database server that stores the primary replica corresponding to the target partition distributes the secondary replica corresponding to the primary replica to the target server.

[0015] This specification provides a copy distribution apparatus, the apparatus comprising:

[0016] The receiving module is used to receive the network latency reported by each database server, wherein the network latency is the network latency between each database server itself and each preset landmark node;

[0017] The determination module is used to determine the target server among the database servers based on the network latency;

[0018] The sending module is used to send the identifier of the target server to the database server that stores the primary replica corresponding to the target partition, so that the database server that stores the primary replica corresponding to the target partition distributes the secondary replica corresponding to the primary replica to the target server.

[0019] This specification provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described copy distribution method.

[0020] This specification provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described copy distribution method.

[0021] The above-described at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects:

[0022] This specification discloses a replica distribution system. In a distributed database system, each database server reports its network latency to each landmark node to a load balancer. The load balancer then selects a target server based on the reported network latency and notifies the database server storing the primary replica corresponding to the target partition to distribute the secondary replicas corresponding to the primary replica to the target server. This method can distribute replicas corresponding to a partition based on the network latency of each database server, enabling flexible deployment of replicas according to the specific network topology of the distributed database system, and effectively improving the query efficiency of the distributed database system. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of this specification and form part of this specification, illustrate exemplary embodiments and are used to explain this specification, but do not constitute an undue limitation thereof. In the drawings:

[0024] Figure 1 This is a schematic diagram of a data query system provided in an embodiment of this specification;

[0025] Figure 2 A flowchart illustrating a copy distribution method provided in an embodiment of this specification;

[0026] Figure 3 A schematic diagram illustrating the determination of the latency representation value from the database server to each landmark node, provided in the embodiments of this specification;

[0027] Figure 4 A schematic diagram of a copy distribution device provided for an embodiment of this specification;

[0028] Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this specification. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0030] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0031] Figure 1 This is a schematic diagram of a data query system provided in an embodiment of this specification. The system includes a load balancing server and several database servers. Figure 1 The data query system shown is essentially a distributed database system. The load balancer is used to distribute user data query requests to a database server, which processes the data query request to retrieve the data to be queried corresponding to the data query request, and returns the data to be queried to the load balancer, which then returns the data to be queried.

[0032] In the embodiments of this specification, the data tables stored in the distributed database system are also partitioned, with each partition corresponding to at least one replica. These replicas include one primary replica and several secondary replicas. Both the primary and secondary replicas contain all the data in the partition, but the different replicas corresponding to the partition are stored on different database servers. For a partition, the database servers on which the replicas corresponding to the partition should be deployed have a significant impact on the query efficiency of the entire distributed database system. Therefore, in the embodiments of this specification, each database server can determine the network latency between itself and the preset landmark nodes in real time and report the network latency to the load balancing server.

[0033] In this specification, the landmark node described in the embodiments is an electronic device that is preset within a certain geographical area and can communicate with each database server via the network. The electronic device can be a database server deployed within the geographical area in a distributed database system, or it can be other servers or terminal devices within the geographical area, as long as it is within the geographical area and can communicate with each database server via the network.

[0034] The network latency between a database server and a landmark node within a geographical range represents the latency at which the database server provides data query services to requesters within that geographical range. Therefore, after receiving the network latency reports from each database server, the load balancer can determine the target server among the database servers based on the network latency between each database server and each landmark node, and send the identifier of the target server to the database server that stores the primary replica corresponding to the target partition.

[0035] The database server that stores the primary copy of the target partition can then distribute the secondary copy corresponding to the primary copy to the target server based on the received identifier of the target server, so that the target server stores the secondary copy.

[0036] Subsequently, when the load balancer receives a data query request for data in the target partition, it can first route the request to the database server storing the primary replica of the target partition (hereinafter referred to as the master node, and the database server storing the secondary replica as the slave node). The master node will attempt to process the query request. If the current load condition of the master node meets the preset conditions, it will directly return the stored primary replica to the load balancer, which will then return the primary replica to the requester. If the current load condition of the master node does not meet the preset conditions, it can route the query request to the slave node with the lowest load, which will then return the stored secondary replica to the load balancer, which will then return the secondary replica to the requester.

[0037] Based on the above Figure 1 The system shown in this specification provides a method for distributing copies, such as... Figure 2 As shown.

[0038] Figure 2 A flowchart of a copy distribution method provided in this specification includes the following steps:

[0039] S200: Receive network latency reported by each database server, wherein the network latency is the network latency between each database server itself and each preset landmark node.

[0040] In the embodiments of this specification, the following are used: Figure 2 The entity performing the replica distribution method shown can be a load balancer server in a distributed database system.

[0041] As mentioned above, in the embodiments of this specification, a topology information management module can be added to each database server. The topology information management module is used to determine the network latency between its own database server and the preset landmark nodes, and report the network latency to the load balancing server.

[0042] Specifically, a topology information management module in a database server can periodically send a message with a first timestamp to a landmark node. The first timestamp corresponds to the time when the topology information management module sends the message to the landmark node. After receiving the message, the landmark node can also return a message with a second timestamp, which corresponds to the time when the landmark node returns the message to the topology information management module. The topology information management module can then determine the network latency between its own database server and the landmark node based on the first and second timestamps.

[0043] S202: Based on the network latency, determine the target server among the database servers.

[0044] In the embodiments of this specification, step S200 is the process by which the load balancing server determines the network topology of the distributed database system. This process can be executed continuously in real time or periodically. Steps S202 to S204 are the distribution and deployment process of replicas corresponding to a partition, which can be triggered by various conditions.

[0045] Specifically, the load balancer server can receive a data query request sent by the requester, and determine the partition where the data to be queried is located based on the data query request, as the target partition. This serves as the trigger for distributing the data to the replicas corresponding to the target partition, and steps S202~S204 are executed.

[0046] In addition, after determining the target partition, the load balancer can also determine whether the current load status of the master node storing the primary replica of the target partition meets the preset conditions. If it meets the conditions, the execution steps S202~S204 can be skipped to distribute and deploy the replica of the target partition. If it does not meet the conditions, the execution steps S202~S204 can be skipped to distribute and deploy the replica of the target partition.

[0047] Of course, after determining the target partition, the load balancer can also determine whether there are any nodes among the master nodes and slave nodes that store the replicas (including both master replicas and slave replicas) corresponding to the target partition whose load conditions do not meet the preset conditions. If there are no nodes, steps S202 to S204 can be skipped to distribute and deploy the replicas of the target partition. If there are nodes, steps S202 to S204 can be skipped to distribute and deploy the replicas of the target partition.

[0048] In the embodiments of this specification, after the load balancing server triggers execution steps S202-S204, it can determine the frequency of requests within a preset geographical range querying data in the target partition, and select the geographical range with the highest frequency as the target geographical range. Then, based on the network latency between each database server and preset landmark nodes within the target geographical range, it determines the target server from among the database servers. Specifically, the load balancing server can determine the database server with the lowest network latency to the landmark nodes within the target geographical range as the target server.

[0049] In the embodiments of this specification, the geographical ranges mentioned above can be divided according to the existing geographical divisions, such as dividing the geographical ranges of provinces, cities or counties into different geographical ranges, or dividing each geographical range according to needs. This specification does not impose any restrictions on this.

[0050] Since a requester needs to send a data query request carrying the key of that target partition to the load balancer when querying data in the target partition, the load balancer can determine the geographical range of the senders (i.e., the requesters) of each data query request based on historically received data query requests. Specifically, the load balancer can determine the geographical range of each sender based on the IP network segment of its Internet Protocol (IP) address and a pre-saved mapping between IP network segments and geographical ranges. Finally, it counts the number of data query requests from each geographical range for the target partition historically, and identifies the geographical range with the highest number of requests as the target geographical range with the highest frequency of data queries for that target partition.

[0051] Once the target geographical area is determined, the load balancer can select the target server with the lowest network latency among the landmark nodes within that geographical area from the database servers.

[0052] S204: Send the identifier of the target server to the database server that stores the primary copy corresponding to the target partition, so that the database server that stores the primary copy corresponding to the target partition distributes the secondary copy corresponding to the primary copy to the target server.

[0053] After the target server is identified among the database servers, the load balancer can send the identifier of the target server to the master node. The master node can then replicate the master replica to obtain the corresponding slave replica, and then distribute the slave replica to the target server. The target server stores the slave replica, and at this point, the target server becomes a slave node.

[0054] It should be noted that the terms "primary replica" and "secondary replica" in the embodiments of this specification are logical concepts. In reality, all secondary replicas are copied from the primary replica, and there is no difference between secondary replicas and primary replicas. The terms "master node" and "secondary node" are logical concepts relative to the replicas corresponding to the same partition. For a given partition, if a database server stores a primary replica of that partition, then that database server is the master node; if a database server stores a secondary replica of that partition, then that database server is the secondary node.

[0055] In this instruction manual Figure 2Steps S200-S204 are executed by a load balancing server. This load balancing server can be an independent server within the distributed database system used for global load balancing of the distributed database system, or it can be a software module located within a database server used for global load balancing of the distributed database system. Regardless of whether the load balancing server is an independent server or a software module within a database server, steps S200-S204 can be implemented by executing a computer program located on the storage medium (including memory, disk, or any other storage medium) of the server or on a processing chip (such as an FPGA or ASIC) connected to the server. This specification does not limit the hardware structure constituting the load balancing server. Furthermore, in this specification, different database servers in the distributed database system can be multiple independent servers, or multiple virtual servers with isolated software environments (such as multiple virtual machines or containers deployed on the same hardware device). It is also possible for a combination of multiple independent servers and multiple virtual servers with isolated software environments to coexist.

[0056] Using the above method, the replicas corresponding to a partition can be distributed to the database server with the lowest network latency in the geographical area where the partition is most frequently queried, based on the network latency of each database server. This not only enables flexible deployment of replicas according to the specific network topology of the distributed database system, but also minimizes network latency when querying data in that partition in subsequent queries, effectively improving the query efficiency of the distributed database system.

[0057] Furthermore, there may be more than one landmark node set within the same geographical area in this specification; therefore, in Figure 2 In step S202 shown, for the target partition, after the load balancer determines the target geographical range with the highest frequency of queries to the target partition, when determining the target server, it can determine the latency characterization value from the database server to the target geographical range based on the network latency between the database server itself and the preset landmark nodes within the target geographical range for each database server. Then, based on the latency characterization values ​​from each database server to the target geographical range, the database server with the lowest latency characterization value is selected from the database servers that do not store the slave replicas corresponding to the primary replicas as the target server.

[0058] Specifically, when determining the latency representation value from a database server to a target geographical area, for each preset landmark node within the target geographical area, the network latency range falling within each preset network latency range can be determined as the target network latency range. The latency representation value corresponding to the target network latency range is then determined as the latency representation value from the database server to the landmark node. Furthermore, based on the latency representation values ​​from the database server to each preset landmark node within the target geographical area, the latency representation value from the database server to the target geographical area is determined. Figure 3 As shown.

[0059] Three landmark nodes, designated as landmark nodes 1 to 3, were set within the target geographic area. The preset network latency range of 0-20ms corresponds to a latency value of 0, the network latency range of 20ms-60ms corresponds to a latency value of 1, and the network latency range of 60ms-120ms corresponds to a latency value of 2. For a database server, assuming its network latency to landmark node 1 is 10ms, to landmark node 2 is 30ms, and to landmark node 3 is 40ms, then the latency values ​​from the database server to landmark nodes 1-3 are 0, 1, and 1, respectively.

[0060] Subsequently, the latency representation value from the database server to the target geographic area can be determined based on the latency representation values ​​of each preset landmark node within the target geographic area. Specifically, the sum or average of the latency representation values ​​from the database server to each landmark node within the target geographic area can be used as the latency representation value from the database server to the target geographic area. Alternatively, the latency representation values ​​from the database server to each landmark node within the target geographic area can be combined into a string according to their priority from high to low, and this string can be used as the latency representation value from the database server to the target geographic area. For example, in the above example, the latency representation values ​​from the database server to landmark nodes 1 to 3 are 0, 1, and 1, respectively. Assuming that landmark nodes 1 to 3 are in the order of priority from high to low as landmark node 1, landmark node 2, and landmark node 3, the combined string is 011. This string 011 is used to represent the latency from the database server to the target geographic area. Figure 3 The latency representation value for the target geographic area is shown. Those skilled in the art should understand that as long as the latency representation value from the database server to the target geographic area can characterize the network latency from the database server to the target geographic area, it is acceptable.

[0061] In the embodiments of this specification, in addition to using the above method to increase the number of slave replicas corresponding to the target partition as the target partition when a data query request for data in a certain partition is received, the load balancer can also adjust the number of slave replicas corresponding to all partitions in real time or periodically based on the network latency reported by each database server.

[0062] Specifically, the load balancer can treat each partition as a target partition, and for each target partition, it uses the same method as above to determine the geographical range where the data in the target partition is queried most frequently, and then uses this range as the target geographical range. Based on the network latency reported by each database server, it identifies the slave nodes among the slave nodes that have stored the slave replicas of the target partition whose latency representation value to the target geographical range is higher than a preset threshold, and uses these slave nodes as database servers to be deleted. The load balancer then notifies the slave replicas corresponding to the target partition to be deleted.

[0063] The aforementioned preset threshold can be a fixed threshold preset as needed, or it can be a dynamic threshold. When the preset threshold is a dynamic threshold, the value of the dynamic threshold can specifically be the minimum value of the latency representation from the database server that does not store any copies of the target partition to the aforementioned target geographical area.

[0064] After the secondary replica corresponding to the target partition is deleted on the database server to be deleted, when the distributed database system receives the next data query request for the target partition, it can use the same method. Figure 2 The method shown deploys a replica of the target partition.

[0065] The above describes a copy distribution system and method provided in the embodiments of this specification. Based on the same idea, this specification also provides corresponding devices, storage media, and electronic devices.

[0066] Figure 4 This is a schematic diagram of a replica distribution device provided in an embodiment of this specification. The device is applied to a load balancing server in a distributed database system; the device includes:

[0067] The receiving module 401 is used to receive the network latency reported by each database server, wherein the network latency is the network latency between each database server itself and each preset landmark node;

[0068] The determination module 402 is used to determine the target server among the database servers based on the network latency;

[0069] The sending module 403 is used to send the identifier of the target server to the database server that stores the primary copy corresponding to the target partition, so that the database server that stores the primary copy corresponding to the target partition distributes the secondary copy corresponding to the primary copy to the target server.

[0070] Optionally, the receiving module 401 is further configured to receive a data query request sent by the requester before the determining module 402 determines the target server among the database servers based on the network latency; and determine the partition where the data to be queried corresponding to the data query request is located as the target partition based on the data query request.

[0071] Optionally, the determining module 402 is further configured to determine, before determining the target server among the database servers based on the network latency, that the current load state of the database server storing the primary copy corresponding to the target partition does not meet a preset condition.

[0072] Optionally, the determining module 402 is specifically used to: determine the frequency of queries made by requesters within a preset geographical range for data in the target partition; take the geographical range with the highest frequency as the target geographical range; and determine the target server among the database servers based on the network latency between each database server and each preset landmark node within the target geographical range.

[0073] Optionally, the determining module 402 is specifically used to: for each database server, determine the latency characterization value from the database server to the target geographical area based on the network latency between the database server itself and each preset landmark node within the target geographical area; and select the database server with the lowest latency characterization value from among the database servers that do not store the slave replicas corresponding to the primary replica, based on the latency characterization values ​​from each database server to the target geographical area, as the target server.

[0074] Optionally, the determining module 402 is specifically used to: for each preset landmark node within the target geographical area, determine the network latency range into which the network latency from the database server to the landmark node falls within each preset network latency range, and use this range as the target network latency range; determine the latency characterization value corresponding to the target network latency range as the latency characterization value from the database server to the landmark node; and determine the latency characterization value from the database server to the target geographical area based on the latency characterization value from the database server to each preset landmark node within the target geographical area.

[0075] Optionally, the determining module 402 is further configured to, based on the network latency reported by each database server, determine, among the database servers that have stored the copy, database servers whose latency representation value to the target geographical area is higher than a preset threshold, and designate them as database servers to be deleted.

[0076] The sending module 403 is also used to notify the database server to be deleted to delete the copy.

[0077] This specification also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can be used to perform the copy distribution method provided above.

[0078] based on Figure 2 The copy distribution method shown in this specification also provides embodiments that... Figure 5 The diagram shows the structure of the electronic device. Figure 5 At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile storage, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile storage into memory and then runs it to implement the aforementioned copy distribution method.

[0079] The above description is merely an embodiment of this specification and is not intended to limit this specification. Various modifications and variations can be made to this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of the claims of this specification.

Claims

1. A copy distribution system, comprising a load balancing server and a plurality of database servers, wherein: any database server is configured to determine network delays between itself and preset landmark nodes, and report the network delays to the load balancing server; the load balancing server is configured to receive the network delays reported by the database servers, determine target servers among the database servers according to the network delays between the database servers and the landmark nodes, and send identities of the target servers to database servers storing master copies corresponding to target partitions; and the database servers storing the master copies corresponding to the target partitions are configured to receive the identities of the target servers, and distribute slave copies corresponding to the master copies to the target servers. The load balancing server is specifically configured to determine frequencies of requests from a geographic range for querying data in a target partition, and determine a target geographic range with the highest frequency; and determine target servers among the database servers according to network delays between the database servers and preset landmark nodes in the target geographic range. 2.A copy distribution method, comprising: receiving network delays reported by database servers, the network delays being network delays between the database servers and preset landmark nodes; determining target servers among the database servers according to the network delays; sending identities of the target servers to database servers storing master copies corresponding to target partitions, so that the database servers storing the master copies distribute slave copies corresponding to the master copies to the target servers; and determining target servers among the database servers according to the network delays, specifically comprising: determining frequencies of requests from a geographic range for querying data in a target partition; determining a target geographic range with the highest frequency; and determining target servers among the database servers according to network delays between the database servers and preset landmark nodes in the target geographic range. 3.The method of claim 2, before determining target servers among the database servers according to the network delays, the method further comprises: receiving a data query request sent by a requester; and determining a partition in which to-be-queried data corresponding to the data query request is located as a target partition according to the data query request. 4.The method of claim 3, before determining target servers among the database servers according to the network delays, the method further comprises: determining that a current load state of a database server storing a master copy corresponding to the target partition does not satisfy a preset condition. 5.The method of claim 2, determining target servers among the database servers according to network delays between the database servers and preset landmark nodes in a target geographic range, specifically comprises: ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ For each database server, a delay representation value of the database server to the target geographical range is determined according to network delays between the database server itself and preset landmark nodes in the target geographical range; According to the delay representation values of the database servers to the target geographical range, a database server with the lowest delay representation value is selected as the target server from the database servers that do not store the slave copy corresponding to the master copy.

6. The method of claim 5, wherein the delay representation value of the database server to the target geographical range is determined according to network delays between the database server itself and preset landmark nodes in the target geographical range, and specifically includes: For each landmark node in the target geographical range, a target network delay range in which the network delay of the database server to the landmark node falls is determined from preset network delay ranges, and a delay representation value corresponding to the target network delay range is determined as the delay representation value of the database server to the landmark node; The delay representation value of the database server to the target geographical range is determined according to the delay representation value of the database server to each landmark node in the target geographical range.

7. The method of claim 5, further comprising: According to the received network delays reported by the database servers, a database server with a delay representation value to the target geographical range higher than a preset threshold is determined as a to-be-deleted database server from the database servers that store the slave copy; The to-be-deleted database server is instructed to delete the slave copy.

8. A replica distribution apparatus, the apparatus comprising: a receiving module configured to receive network delays reported by database servers, the network delays being network delays between the database servers themselves and preset landmark nodes; a determining module configured to determine a target server from the database servers according to the network delays; a sending module configured to send an identifier of the target server to a database server storing a master copy corresponding to a target partition, so that the database server storing the master copy corresponding to the target partition distributes a slave copy corresponding to the master copy to the target server; The determining module is specifically configured to determine a frequency at which a requester located in a preset geographical range queries data in a target partition, and determine a geographical range with the highest frequency as a target geographical range; The target server is determined from the database servers according to network delays between the database servers themselves and preset landmark nodes in the target geographical range.

9. A computer-readable storage medium, the storage medium storing a computer program, the computer program being executed by a processor to implement the method of any one of claims 2-7.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of claims 2-7 when executing the program.

Images