CN106817408B - Distributed server cluster scheduling method and device - Google Patents

Abstract

The invention discloses a distributed server cluster scheduling method and device, and relates to the technical field of operation platform management. The method includes: a main server receives an execution message sent by a network server via a message queue server, wherein the execution message includes the correspondence between each client server and the server. The master server obtains the status of each slave server in the slave server cluster, and the slave server cluster includes at least two slave servers; the master server calls the client server’s The maximum threshold and the state of each slave server determine the slave server to be called; the master server distributes the pending tasks corresponding to each client server in the execution message to the slave server to be called. In the embodiment of the present invention, by calling the slave server in the server cluster, the task is quickly delivered to the client corresponding to the task to execute the task, so that the utilization of the client is maximized and high availability is achieved.

Description

Distributed server cluster scheduling method and device

technical field

The invention relates to the technical field of operation platform management, and in particular, to a method and device for scheduling distributed server clusters.

Background technique

Distributed database system has become an important field of information processing and is developing rapidly. It has the following advantages: 1. It can solve the problem that organizations are scattered and data needs to be interconnected. For example, in a banking system, the head office and branches are located in different cities or in various regions of the city. In terms of business, they need to process their own data, and also need to exchange and process each other, which requires a distributed system. 2. If an organization needs to add new relatively autonomous organizational units to expand the organization, the distributed database system can be expanded with the least impact on the current organization. 3. The need for load balancing. The decomposition of data is used to maximize local application, which minimizes the mutual interference between processors. The load is shared among processors to avoid critical bottlenecks. 4. When there are several database systems in the existing organization, and the necessity of realizing the global application increases, the distributed database system can be composed of these databases from the bottom up. 5. The probability of failure of a distributed database system of equal scale is not lower than that of a centralized database system, but since the impact of its failure is limited to local data applications, its reliability is relatively high in terms of the entire system .

With the rapid growth of enterprise business, the number of servers and the maintenance requirements for servers are increasing day by day. However, in the prior art, a single-node server is used as a task distribution platform, which lacks a high-availability mechanism, slows task execution, and cannot manage computer clusters in a timely and efficient manner.

Therefore, to sum up, the prior art cannot provide an effective and efficient method for assigning tasks.

SUMMARY OF THE INVENTION

The present invention provides a distributed server cluster scheduling method and device, which are used to solve the problem that an effective and efficient task dispatching method cannot be provided in the prior art.

An embodiment of the present invention provides a distributed server cluster scheduling method, and the method includes:

The main server receives the execution message sent by the network server via the message queue server, wherein the execution message includes tasks to be processed corresponding to each client server;

The master server obtains the status of each slave server in the slave server cluster, wherein the slave server cluster includes at least two slave servers;

The master server determines the slave server to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the state of each slave server;

The master server distributes the tasks to be processed corresponding to each client server in the execution message to the slave servers that need to be called.

In this embodiment of the present invention, after the master server receives the tasks to be processed, according to the status of the slave servers in the slave server cluster, the number of tasks, the maximum threshold of clients that each slave server can call, and the current running state of each slave server , determine the slave servers that need to be called, and issue the task to these slave servers. In the embodiment of the present invention, by calling the slave server in the server cluster, the task can be quickly dispatched to the client corresponding to the task to execute the task, so that the utilization of the client is maximized, and the slave server in the slave server cluster can fail. At the same time, the task can also be smoothly delivered to the client, achieving high availability.

Further, the master server distributes the pending tasks corresponding to each client server in the execution message to the slave servers that need to be called, including:

The master server divides the tasks to be processed corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and sends the task list to the corresponding slave servers that need to be called. server;

After the master server distributes the pending tasks corresponding to each client server in the execution message to the slave servers to be called, the method further includes:

For each slave server to be called, the slave server to be called calls the corresponding client server according to the received task list; the slave server to be called receives the task processing result of the client server, and the The task processing result is sent to the network server through the message queue server.

In the embodiment of the present invention, the master server issues the acquired task to the slave server of the client corresponding to the calling task, so that the slave server can determine which clients to call to execute the task.

Further, the master server obtains the status of each slave server in the slave server cluster, including:

The management server obtains the status of each slave server in the slave server cluster;

The management server synchronizes the state of each slave server to at least one standby management server;

The main server determines whether the running state of the management server is normal;

If the master server determines that the state of the management server is abnormal, call any one of the at least one standby management server to obtain the state of each slave server in the slave server cluster.

In the embodiment of the present invention, the management server can synchronize the status of each slave server to the standby management server, so that when the management server fails, the status of each slave server can be obtained in time, thereby realizing high availability.

Further, the management server obtains the status of each slave server in the slave server cluster, including:

For any slave server in the slave server cluster, if the management server cannot obtain the heartbeat detection information of the slave server within a set time, it is determined that the state of the slave server is abnormal.

In the embodiment of the present invention, each slave server is connected to the management server through heartbeat detection information. If the management server cannot receive the heartbeat detection message of a certain slave server within a set time, it is considered that the slave server cannot be used, and then the When the master server queries the status, the slave server's status information is no longer available.

Further, the method also includes:

If the master server determines that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, it adds a new slave server, and the newly added slave server is registered in the management server.

In the embodiment of the present invention, when it is determined that the existing slave server cluster can no longer meet the requirements of task processing, the slave server is dynamically added to achieve scalability.

The present invention also provides a distributed server cluster scheduling device, comprising:

a receiving unit, configured to receive an execution message sent by the network server via the message queue server, wherein the execution message includes tasks to be processed corresponding to each client server;

an obtaining unit, configured to obtain the status of each slave server in the slave server cluster, wherein the slave server cluster includes at least two slave servers;

a determining unit, configured to determine the slave server to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the state of each of the slave servers;

A sending unit, configured to distribute the tasks to be processed corresponding to each client server in the execution message to the slave server that needs to be called.

In the embodiment of the present invention, by calling the slave server in the server cluster, the task can be quickly dispatched to the client corresponding to the task to execute the task, so that the utilization of the client is maximized, and the slave server in the slave server cluster can fail. At the same time, the task can also be smoothly delivered to the client, achieving high availability.

Further, the determining unit is specifically used for:

Divide the tasks to be processed corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and send the task list to the corresponding slave servers that need to be called;

Invoke the corresponding client server according to the received task list; receive the task processing result of the client server, and send the task processing result to the network server through the message queue server.

Further, the acquisition unit is specifically used for:

Get the status of each slave server in the slave server cluster;

Synchronize the state of each slave server to at least one standby management server;

determining whether the running state of the management server is normal;

If it is determined that the state of the management server is abnormal, any one of the at least one standby management server is called to obtain the state of each slave server in the slave server cluster.

Further, the acquisition unit is specifically used for:

For any slave server in the slave server cluster, if the heartbeat detection information of the slave server cannot be acquired within a set time, it is determined that the state of the slave server is abnormal.

Further, the device also includes:

A dynamic adding unit, configured to add a slave server if it is determined that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, and the newly added slave server is registered in the management server.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic structural diagram of a distributed server cluster scheduling system according to an embodiment of the present invention;

2 is a schematic diagram of a message queue in a message queue server according to an embodiment of the present invention;

3 is a schematic structural diagram of a distributed server cluster scheduling apparatus according to an embodiment of the present invention;

FIG. 4 is a schematic flowchart of a method for scheduling a distributed server cluster according to an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. . Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The present invention provides a distributed server cluster scheduling method, as shown in FIG. 1 , including:

Master server 101 , network server 102 , message queue server 103 , slave server 104 , client server 105 and management server 106 .

The network server 102 is used to manage the system, obtain information and feed it back to the upper layer, or obtain commands from the upper layer and execute the commands in the system. In this embodiment of the present invention, the network server 102 may formulate tasks, make the tasks into a task list, and send the tasks to the client server 105 for execution, wherein the tasks in the task list have a corresponding relationship with the client server 105 .

For example, in this embodiment of the present invention, the task list formulated by the network server 102 is as shown in Table 1, specifically, task 1 is issued to client 1, task 2 is issued to client 2, and task 3 is issued to Task side 3.

Task content Client List task content 1 Client 1 task content 2 Client 2 task content 3 Client 3

Table 1: Task List

In the embodiment of the present invention, the main server 101 encapsulates the task list into an execution message, and sends it to the message queue server 103. In the embodiment of the present invention, optionally, as shown in FIG. 2, the message queue server 103 According to the order in which the messages are received, multiple messages are arranged in order. Assuming that the execution message is the third message in sequence, after the message queue server 103 processes the first two messages, the execution message in the embodiment of the present invention is processed. .

Optionally, in this embodiment of the present invention, the message queue server 103 subscribes to the message of the network server 102 , that is, when there is a new message in the network server 102 , the message queue server 103 can obtain the new message.

After receiving the execution message sent by the message queue server 103, the main server 101 determines the task list in the execution message according to the execution message. For example, in this embodiment of the present invention, the main server 101 receives the tasks in Table 1 in the foregoing embodiment. list, the master server 101 determines the slave server 104 in the slave server cluster to be called according to the task list.

In this embodiment of the present invention, the slave server cluster is a server cluster with a large number of slave servers 104, and the slave servers 104 in the slave server cluster can call the client server 105.

Optionally, in this embodiment of the present invention, the number of each slave server 104 in the slave server cluster that can call the client server 105 has a maximum value, which can be set to m, that is, one slave server 104 can call m clients. End server 105.

In this embodiment of the present invention, before the master server 101 calls the slave server 104, it first needs to determine the state of each slave server 104 in the slave server cluster. Optionally, in this embodiment of the present invention, the state of the slave server 104 includes an abnormal state and normal state.

In this embodiment of the present invention, the master server 101 can obtain the status of the slave servers 104 through the management server 106. When a slave server cluster is established, each slave server 104 in the slave server cluster will send a connection request to the management server 106, and the management The server 106 can determine that the slave server 104 has joined the slave server cluster according to the connection request.

After the connection between the slave server 104 and the management server 106 is established, the heartbeat detection message is used to determine whether the slave server 104 is abnormal. Optionally, if the management server 106 does not receive the heartbeat detection message sent from the The status given to the slave server 104 is considered abnormal.

Optionally, in this embodiment of the present invention, the management server 106 stores a state information table of all slave servers 104 registered in the management server 105, and the table is updated periodically. For example, as shown in Table 2, if N slave servers 104 are registered in the management server 106, and the state information table of the slave server 104 is updated once in the management server 106, the state information table stores two states of each slave server. .

slave server state 1 state 2 slave server 1 normal normal slave 2 normal abnormal ... ... ... Slave N normal normal

Table 2: Status information table in the management server

Optionally, in this embodiment of the present invention, the management server 106 also saves two status information tables, one is an information table that the slave server 104 is in a normal state, as shown in Table 3, and the other is an abnormal state of the slave server 104. information table, as shown in Table 4.

slave server state slave server 1 normal slave server 3 normal ... ... slave server N1 normal

Table 3: Status information table in the management server

slave server state slave 2 abnormal slave server 4 abnormal ... ... slave server N2 abnormal

Table 4: Status information table in the management server

All stored in Table 3 are the latest lists of slave servers 104 whose status is normal, such as slave server 1, slave server 3, etc. in Table 3; optionally, the identification information of slave server 104 is stored, which can be slave The hardware address of the server 104 or other identifiers that can characterize the uniqueness of the slave server 104 .

Table 4 stores all the latest lists of slave servers 104 whose status is abnormal, such as slave server 2, slave server 4, etc. in Table 4; optionally, it stores the identification information of slave server 104, which can be slave server The hardware address of 104 or other identifiers that can characterize the uniqueness of the slave server 104 .

Optionally, in this embodiment of the present invention, the management server 106 can also obtain the working status of each slave server list 104 in the list of slave servers 104 whose status is normal. Whether the slave server 104 has called the client server 105, and called several client servers 105, as shown in Table 5, the management server 106 also saves the status of each slave server 104 in the normal list of slave servers 104. The number of client server 105 calls.

slave server state used slave server 1 normal 2 units have been called slave server 3 normal 0 units have been called slave server 4 normal m stations have been called ... ... ...

Table 5: The working status table of the slave server saved in the management server

The master server 101 can determine whether the slave server 104 can be called according to the working status table in Table 5, and can also use the slave server 104 to call how many slave servers. For example, in this embodiment of the present invention, the slave server 104 calls the client server. The maximum number of 105 is m, that is to say, as from server 4 in Table 5, if m client servers 105 have been called, the client server 105 cannot be called any more. The slave server 1 has already called 2 client servers 105, then m-2 client servers 105 can also be called.

Optionally, in this embodiment of the present invention, the management server 106 replicates the underlying data of all the acquired information lists of the slave server 104, and replicates it to at least one standby management server 106, and the master server 101 obtains it from the management server 106. When the information of the slave server 104 is used, it is necessary to first determine whether the management server 106 is abnormal.

The master server 101 determines which slave servers to call in the slave server cluster according to the number of client servers 105 that need to perform tasks in the received task list, the current state of the slave server 104 and the maximum threshold for calling the client server 105 from the slave server 104 104.

In this embodiment of the present invention, assuming that the number of client servers 105 that need to perform tasks is n, there are 20 slave servers 104 in the queried slave server cluster, of which 4 slave servers 104 have called 0, and there are 10 slave servers 104 The slave server 104 has called m units, and there are 6 slave servers 104 that have called 2 units. The master server 101 first determines whether the number of client servers 105 that need to perform tasks satisfies Formula 1:

n≤(m-2)×6+4m (Formula 1)

If formula 1 is satisfied, for example, n=(m-2)×6+4m, all slave servers 104 except 10 slave servers 104 that have called m client servers 105 in the slave service cluster are called, and 4 slave servers 104 have been called. Each of the slave servers 104 calling 0 client servers 105 calls m client servers 105, and each of the 6 slave servers 104 that have called 2 client servers 105 calls m-2 client servers 105 each.

The master server 101 distributes the n client servers 105 that need to perform tasks in the task list to each of the above slave servers 104, that is, randomly assigns the n client servers 105 to 10m-12 slave servers 104, each slave server 104. After receiving the client server 105 to be called, the server 104 calls the corresponding client server 105 to execute the task in the task list.

In this embodiment of the present invention, the task processing result of the called client server 105 after processing the task is received from the server 104 , and the task processing result is sent to the message queue server 103 , and the message queue server 103 sends it to the network server 102 .

Optionally, in this embodiment of the present invention, when it is determined that formula 1 is not satisfied, the number of slave servers 104 in the slave server cluster may be dynamically increased. server 104 and register the new slave server 104 in the management server 106 .

Based on the same concept, an embodiment of the present invention also provides a distributed server cluster scheduling device, as shown in FIG. 3 , including:

A receiving unit 301, configured to receive an execution message sent by a network server via a message queue server, wherein the execution message includes tasks to be processed corresponding to each client server;

an obtaining unit 302, configured to obtain the status of each slave server in the slave server cluster, wherein the slave server cluster includes at least two slave servers;

Determining unit 303, configured to determine the slave server that needs to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the state of each of the slave servers;

The sending unit 304 is configured to distribute to-be-processed tasks corresponding to each client server in the execution message to the slave server that needs to be called.

Further, the determining unit 303 is specifically configured to:

Divide the tasks to be processed corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and send the task list to the corresponding slave servers that need to be called;

Invoke the corresponding client server according to the received task list; receive the task processing result of the client server, and send the task processing result to the network server through the message queue server.

Further, the obtaining unit 302 is specifically used for:

Get the status of each slave server in the slave server cluster;

Synchronize the state of each slave server to at least one standby management server;

determining whether the running state of the management server is normal;

If it is determined that the state of the management server is abnormal, any one of the at least one standby management server is called to obtain the state of each slave server in the slave server cluster.

Further, the obtaining unit 302 is specifically used for:

For any slave server in the slave server cluster, if the heartbeat detection information of the slave server cannot be acquired within a set time, it is determined that the state of the slave server is abnormal.

Further, the device also includes:

The dynamic adding unit 305 is configured to add a slave server if it is determined that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, and the newly added slave server is registered in the management server.

Based on the same concept, the present invention also provides a distributed server cluster scheduling method, as shown in FIG. 4 , including:

Step 401, the main server receives the execution message sent by the network server via the message queue server, wherein the execution message includes the tasks to be processed corresponding to each client server;

Step 402, the master server obtains the status of each slave server in the slave server cluster, wherein the slave server cluster includes at least two slave servers;

Step 403, the master server determines the slave server to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the state of each slave server;

Step 404, the master server distributes the tasks to be processed corresponding to each client server in the execution message to the slave servers that need to be called.

Further, the master server distributes the pending tasks corresponding to each client server in the execution message to the slave servers that need to be called, including:

The master server divides the tasks to be processed corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and sends the task list to the corresponding slave servers that need to be called. server;

After the master server distributes the pending tasks corresponding to each client server in the execution message to the slave servers to be called, the method further includes:

For each slave server to be called, the slave server to be called calls the corresponding client server according to the received task list; the slave server to be called receives the task processing result of the client server, and the The task processing result is sent to the network server through the message queue server.

Further, the master server obtains the status of each slave server in the slave server cluster, including:

The management server obtains the status of each slave server in the slave server cluster;

The management server synchronizes the state of each slave server to at least one standby management server;

The main server determines whether the running state of the management server is normal;

If the master server determines that the state of the management server is abnormal, call any one of the at least one standby management server to obtain the state of each slave server in the slave server cluster.

Further, the management server obtains the status of each slave server in the slave server cluster, including:

For any slave server in the slave server cluster, if the management server cannot obtain the heartbeat detection information of the slave server within a set time, it is determined that the state of the slave server is abnormal.

Further, the method also includes:

If the master server determines that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, it adds a new slave server, and the newly added slave server is registered in the management server.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device produce Means for implementing the functions specified in a flow or flow of a flowchart and/or a block or blocks of a block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions The apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that The instructions provide steps for implementing the functions specified in the flow or blocks of the flowcharts and/or the block or blocks of the block diagrams.

Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiment and all changes and modifications that fall within the scope of the present invention.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention. Thus, provided that these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims (10)

1. A distributed server cluster scheduling method, wherein the method comprises: The main server receives the execution message sent by the network server via the message queue server, wherein the execution message includes tasks to be processed corresponding to each client server; The master server obtains the status of each slave server in the slave server cluster, wherein the slave server cluster includes at least two slave servers; The master server determines the slave server that needs to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the status of each slave server, wherein if the following formula is satisfied, then The master server calls the slave server corresponding to the client server: n≤(m-2)×6+4m Wherein, n is the number of client servers that need to be called, and m is the maximum number of the client servers that the slave server calls; The master server distributes the tasks to be processed corresponding to each client server in the execution message to the slave servers that need to be called. 2. The method according to claim 1, wherein the master server distributes the tasks to be processed corresponding to each client server in the execution message to the slave servers that need to be called, comprising: The master server divides the tasks to be processed corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and sends the task list to the corresponding slave servers that need to be called. server; After the master server distributes the pending tasks corresponding to each client server in the execution message to the slave servers to be called, the method further includes: For each slave server to be called, the slave server to be called calls the corresponding client server according to the received task list; the slave server to be called receives the task processing result of the client server, and the The task processing result is sent to the network server through the message queue server. 3. The method according to claim 1, wherein the master server obtains the status of each slave server in the slave server cluster, comprising: The management server obtains the status of each slave server in the slave server cluster; The management server synchronizes the state of each slave server to at least one standby management server; The main server determines whether the running state of the management server is normal; If the master server determines that the state of the management server is abnormal, call any one of the at least one standby management server to obtain the state of each slave server in the slave server cluster. 4. The method according to claim 3, wherein the management server obtains the status of each slave server in the slave server cluster, comprising: For any slave server in the slave server cluster, if the management server cannot obtain the heartbeat detection information of the slave server within a set time, it is determined that the state of the slave server is abnormal. 5. The method according to claim 3, wherein the method further comprises: If the master server determines that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, it adds a new slave server, and the newly added slave server is registered in the management server. 6. A distributed server cluster scheduling device, comprising: a receiving unit, configured to receive, through the main server, an execution message sent by the network server via the message queue server, wherein the execution message includes tasks to be processed corresponding to each client server; an obtaining unit, configured to obtain the status of each slave server in the slave server cluster through the master server, wherein the slave server cluster includes at least two slave servers; a determining unit, configured to determine, by the master server, the slave server to be called according to the number of client servers of the task to be processed, the maximum threshold for calling the client server from the slave server, and the status of each slave server, wherein If the following formula is satisfied, the determining unit calls the slave server corresponding to the client server: n≤(m-2)×6+4m Wherein, n is the number of client servers that need to be called, and m is the maximum number of the client servers that the slave server calls; A sending unit, configured to distribute, through the master server, the tasks to be processed corresponding to each client server in the execution message to the slave servers that need to be called. 7. The device according to claim 6, wherein the determining unit is specifically configured to: The master server divides the pending tasks corresponding to the client servers in the execution message into task lists corresponding to the slave servers that need to be called, and sends the task list to the corresponding slave servers that need to be called. from the server; Call the corresponding client server through the main server according to the received task list; receive the task processing result of the client server through the main server, and send the task processing result to the message queue server through the message queue server. Network Server. 8. The device according to claim 6, wherein the acquiring unit is specifically configured to: Obtain the status of each slave server in the slave server cluster through the management server; Synchronizing the state of each slave server to at least one standby management server through the management server; Determine by the main server whether the running state of the management server is normal; If it is determined by the master server that the state of the management server is abnormal, any one of the at least one standby management server is called to obtain the state of each slave server in the slave server cluster. 9. The device according to claim 8, wherein the acquiring unit is specifically configured to: For any slave server in the slave server cluster, if the management server cannot obtain the heartbeat detection information of the slave server within a set time, it is determined that the state of the slave server is abnormal. 10. The apparatus of claim 8, wherein the apparatus further comprises: A dynamic adding unit, configured to add a new slave server if it is determined by the master server that the usage rate of each slave server in the slave server cluster is equal to the maximum threshold, and the newly added slave server is registered in the in the management server.

Images