JP4900358B2 - Server management system, server management method, and server management program - Google Patents

Description

  The present invention relates to a server management system, a server management method, and a server management program for monitoring the operating status of a plurality of servers.

  In a place where a plurality of servers are provided such as a data center, a server management system for comprehensively managing the plurality of servers is constructed. An ordinary server management system is configured by connecting a plurality of servers and a server management apparatus to each other via a line, and the server management apparatus communicates with each server individually at a fixed period, and each server is activated or deactivated. Monitoring and performance monitoring were performed. That is, the server transmits a life notification as life and death monitoring to the server management apparatus at a fixed period, and transmits performance information such as its own load state as performance monitoring.

  Information transmission / reception methods between the server and the server management device include a method in which each server transmits a transmission right (token) and only the server having the transmission right transmits performance information to the server management device. There is a method of transmitting performance information to the server management apparatus based on each determination.

  When such a method is applied to a large-scale system consisting of thousands of servers or more, the former method takes time until the transmission right is distributed to all servers. The communication cycle for each server of the apparatus becomes longer, and it becomes impossible to detect an abnormality of the server at an early stage. Even with the latter method, the server management device has a limited number of communications that can be processed within a certain time, and it takes a considerable amount of time for the server management device to receive information from all of the large number of servers. In the same way as the former method, it is necessary to increase the communication interval for each server.

  In this regard, related techniques are disclosed in Patent Documents 1 and 2. Patent Document 1 discloses a computer monitoring method in which servers monitor each other and collect failure detection results on a route using a tree-structured logical network. In this method, the servers perform mutual monitoring, so the communication of the management device is only at the time of acquisition of the failure detection result. Even if the number of servers becomes enormous, the monitoring cycle for each server remains short. It was possible to detect server failure.

  Further, in Patent Document 2, servers (mutual monitoring devices) monitor each other, notify a management server of a failure detection result, and a server that detects the failure is a monitoring target server of the failed server. A system for taking over is disclosed.

JP 2003-271471 A JP 2007-114941 A

  However, in the related technologies disclosed in Patent Documents 1 and 2 described above, the servers can monitor each other at an effective cycle by distributing the server monitoring processing by monitoring the operating status between the servers. However, the management console (management server), which should manage all the servers in an integrated manner, can detect the occurrence of a server failure and perform performance monitoring to manage the performance information (operation status) of each server collectively. Could not do. Obtaining and managing the performance information of each server periodically is indispensable for system operation management, and there is a disadvantage that load distribution control cannot be performed unless performance monitoring of each server can be performed.

  Even if the related technique disclosed in Patent Document 1 is applied to performance monitoring, a tree-structure topology is adopted for collecting performance information, so that servers belonging to groups near the root always have the performance of a large number of servers. Information will be handled. A server in a data center normally operates as a business system. If the load is always concentrated on a specific server, the business system is hindered.

  Accordingly, the present invention provides a server monitoring system that improves the disadvantages of the related art and can effectively monitor each server by collecting performance information of all servers at an appropriate period for a large-scale system having a plurality of servers. An object of the present invention is to provide a server monitoring method and a server monitoring program.

To achieve the above object, the server monitoring system of the present invention, a plurality of servers and server management device and a server monitoring system formed by the line connected to one another, each server, or de Brew Yan graph structure with multiple servers Kautsugurafu monitoring topology information storing means for storing information relating to the own server or specified in the server management device as a connection destination or other servers predetermined so as to constitute the monitoring topology structure, this stored information Such other server as a monitoring target server, monitoring means for repeatedly acquiring the performance information of the monitoring target server at a preset time interval, and monitoring information for accumulating the acquired performance information of the monitoring target server as monitoring information The monitoring information received from the storage means and the server management device is received and this inquiry is monitored When a query transferred from another server is received, this query is relayed and sent to the monitoring target server, and the monitoring information storage means responds to a query from the server management device or another server. The server management apparatus includes a monitoring information inquiry reply unit that replies with the stored monitoring information and relays the monitoring information when the monitoring information returned from the monitoring target server is received and transmits the monitoring information to the inquiry transmission source. Based on the monitoring information related to all the servers recently collected at a preset time interval, the server with the lowest load is sequentially selected from a plurality of servers, and the inquiry of the monitoring information is transmitted to the selected server. Monitoring information collecting means for receiving monitoring information related to all servers returned to the server, and a monitoring information management unit for managing the received monitoring information related to all servers. Together and means, monitoring information inquiry reply unit of each server, when receiving the same query from different sources, second and subsequent in place of the monitoring information to the transmission source of the query received secondly or later It is characterized in that it has a function of returning a notification notifying that it is an inquiry .

Also, the server monitoring method of the present invention is a server monitoring system in which a plurality of servers and a server management apparatus are connected to each other via a line, and each server forms a monitoring topology of a Doubling graph structure with a plurality of servers. The server management device is designated as the connection destination of its own server, or the performance information of the monitored server is repeatedly acquired at a preset time interval using another predetermined server as the monitored server. The performance information is stored as monitoring information in the monitoring information storage means provided in advance, and the server management device loads the most load from a plurality of servers at a preset time interval based on the monitoring information relating to all recently collected servers. sequentially selects a small server sends a query monitoring information to the selected server, receives the query from the server management apparatus Server sends the inquiry to the two monitored server which is set in advance, the query is transmitted along the monitoring topology to each server relays all servers, the query in this monitoring information along the monitoring topology Each server relays and transmits to all servers, and the server that has received the inquiry from the server monitoring device or another server returns the monitoring information stored in the monitoring information storage means to the transmission source of the inquiry. monitoring information transmitted each server and relays along the reverse monitoring topology according to the server management device, receives monitoring information from all servers from directly sending the previous server queries the monitoring information with The server that received the query from two other servers returns a notification to the sender of the second query that it was the second time, and both of the two monitored servers The server that has received the reply to the inquiry returns the monitoring information stored in the monitoring information storage means to the transmission source of the first inquiry, and the server that has received the monitoring information reply from the monitoring target server The monitoring information stored in the monitoring information storage means of the own server is added to the monitoring information from the server and transmitted to the transmission source of the first inquiry, and the monitoring information related to each server is sent along the reverse direction of the monitoring topology. Receiving from the server to which the server management device sent the monitoring information query directly, the monitoring information related to all servers that the server relayed and transmitted while accumulating, and accumulated monitoring information related to each server. Features.

The server monitoring program according to the present invention is provided in each server in a server monitoring system in which a plurality of servers and a server management apparatus are connected to each other in a line, and the plurality of servers constitute a monitoring topology having a Doubling graph structure. The server management device is designated as the connection destination of its own server, or the performance information of the monitored server is repeatedly acquired at a preset time interval using another predetermined server as the monitored server. A monitoring function for storing performance information as monitoring information in a monitoring information storage means prepared in advance, and a query for monitoring information received from the server management apparatus, and sending this query to two preset servers to be monitored. the relays the inquiry along when monitoring topology received from other servers transmitted to the managed server The monitoring information inquiry function and the monitoring information stored in the monitoring information storage means in response to an inquiry from the server management device or other server are returned to the inquiry transmission source, and the monitoring information returned from the monitoring target server is received. In this case, the monitoring information reply function for relaying the monitoring information and transmitting it to the inquiry transmission source is executed by the computer, and the monitoring information reply function can be used for the second and subsequent times when an equivalent inquiry is received from different transmission sources. It is a function that returns a notification notifying the second and subsequent inquiries to the sender of the received inquiry instead of the monitoring information .

  Since the present invention is configured and functions as described above, this constitutes a monitoring topology having a specific structure made up of a plurality of servers, and the servers monitor the servers according to their connection relations. The server management device sends a monitoring information query to any one of a plurality of servers, and the query is sequentially propagated to all servers along a monitoring topology composed of a plurality of servers. Monitoring information returned from the server is relayed to each server and sent back to the server management device via the server directly inquired from the server management device, so that server management can be performed without concentrating the load on a specific server. The load on the communication of the apparatus is reduced, and performance information of a large number of servers can be collected and managed at an appropriate cycle.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram showing the configuration of the server monitoring system of this embodiment. As shown in FIG. 1, the server monitoring system of this embodiment is configured such that a server management apparatus M and n (n is a natural number of 4 or more) machines A1, A2, A3 to An are mutually connected via a network L. It is a connected configuration. The servers A1 to An are information processing apparatuses, which are business servers in which application software is installed and a business system is configured by one or a plurality of servers. The business server is, for example, a Web server, an application server, a database server, or the like.

  FIG. 2 is a functional block diagram illustrating the internal configuration of the servers A1 to An and the server management apparatus M in the present embodiment.

  Here, since the servers A1 to An of the present embodiment have the same configuration, FIG. 2 shows only the configuration of the server A1 among the servers A1 to An for convenience. Further, the server A2 and the server An shown in FIG. 2 are not shown in the configuration, and the configurations of the server A2 and the server An are also equivalent to the server A1.

  As shown in FIG. 2, the servers A <b> 1 to An in this embodiment are the monitoring IDs assigned so that the servers A <b> 1 to An constitute a monitoring topology having a specific graph structure, and other connection destinations of the own server. The monitoring topology information storage unit 11 for storing server information and the other server of the connection destination related to the stored information as the monitoring target server repeatedly acquire performance information of the monitoring target server at a preset time interval. A monitoring unit 12 for determining whether or not an abnormality has occurred in the monitoring target server based on the acquired performance information and notifying the server management apparatus M of the abnormality, and the acquired monitoring target Monitoring information storage means 13 for storing server performance information as monitoring information, and monitored means 14 for transmitting the performance information of the own server to other servers are provided.

  Furthermore, the servers A1 to An receive a monitoring information query from the server management apparatus M, send this query to the monitoring target server, and if this query is received from another server, relay this query and monitor it. When the inquiry is received from the server management apparatus or another server, the monitoring information stored in the monitoring information storage means 13 is returned to the inquiry transmission source and returned from the monitoring target server. When monitoring information is received, monitoring information inquiry reply means 15 having a function of relaying the monitoring information and transmitting it to the inquiry source, and a server serving as a monitoring ID of the own device and a connection destination of the own device from the server management apparatus M The monitoring topology information updating unit 16 updates the information stored in the monitoring topology information storage unit 11 when the change is notified.

  The server management apparatus M sequentially selects servers that meet the specified conditions from a plurality of servers A1 to An at a preset time interval, and sends a monitoring information query to the selected servers. Monitoring information collecting means 21 for receiving monitoring information relating to all servers A1 to An returned from the server, monitoring information managing means 22 for managing the received monitoring information relating to all servers A1 to An, and each server A1 An anomaly detection means 23 for receiving a notification of an abnormality of the monitored server from An, and a reconfiguration of a monitoring topology having a specific graph structure on a plurality of servers excluding the server on which the abnormality has occurred. And a monitoring topology configuration unit 24 that transmits information of the connection destination server after the change to the server in which the setting of the connection destination server has been changed.

  Here, the monitoring topology of a specific graph structure is a logical topology that connects the servers A1 to An, and for example, a de Bruijn graph structure or a Kautz graph structure is suitable as the specific graph structure. . The de Bruijn graph and Kautz graph structures have a small distance between the two most distant nodes, and the number of nodes connected to each node is small, so the topology of the Bruijn graph and Kautz graph structure improves communication efficiency. And a topology that can reduce the burden on the server. Here, as a reference example, a de Bruijn graph structure in the case of eight nodes is shown in FIG.

  The monitoring topology information storage means 11 in the servers A1 to An is the connection destination of the own server and the monitoring ID assigned to each server so that only a plurality of servers A1 to An constitute a monitoring topology having a specific graph structure. Server information is stored.

  The monitoring unit 12 has a function of acquiring performance information of the monitoring target server from the monitored unit 14 provided in the monitoring target server at a constant time interval or a random time interval and storing the performance information in the monitoring information storage unit 13. . As a method for acquiring the performance information of the monitoring target server, a transmission request is repeatedly transmitted to the monitoring target server at a constant time interval or a random time interval, and the monitored means 14 of the monitoring target server responds to this request. Adopt a method to acquire performance information. Here, in FIG. 2, the server A1 is the server to be monitored by the server A1, and the server An is the server having the server A1 as the server to be monitored.

  Further, the monitoring unit 12 determines that the numerical information indicated in the performance information from the monitoring target server is abnormal when the response from the monitoring target server has not been replied to the performance information transmission request for a certain period of time. If the value is a value, it is determined that an abnormality has occurred in the monitoring target server, and a function of notifying the server management apparatus M that an abnormality has occurred in the monitoring target server is provided.

  Here, the performance information of the servers A1 to An is, for example, measured numerical information such as the current CPU usage rate, memory consumption, HDD free capacity, or information on the server status determined based on the information. Yes, the monitoring unit 12 may determine that an abnormality has occurred in the monitoring target server when detecting that the CPU usage rate remains 100% for a long time.

  When the monitored means 14 in the servers A1 to An receives a performance information transmission request from another server, the monitored means 14 includes an internal measurement such as CPU usage rate, memory consumption, HDD free space, etc. A function of returning performance information as information e to the other server is provided. As described above, the servers A1 to An each include the monitoring unit 12 and the monitored unit 14, and can monitor the operations of the servers.

  The monitoring information inquiry reply means 15 receives the inquiry of the monitoring information from the server management apparatus M, transfers this inquiry to the monitoring target server, and receives the inquiry transferred from another server, this inquiry is sent to the monitoring target server. A function to send, a function to send back a notification notifying the second and subsequent inquiries to the transmission source of the second and subsequent inquiries when receiving a query from different transmission sources, and from all monitored servers When the reply to the inquiry is confirmed, the monitoring information stored in the monitoring information storage means 13 is returned to the transmission source of the first inquiry, and when the reply from the monitoring target server is the monitoring information, the monitoring information is A function of adding to the monitoring information in the monitoring information storage means 13 and transmitting it to the transmission source of the first inquiry.

  Here, “some reply to the inquiry from the monitoring target server” is a notification that the inquiry has been made for the second time or later, or monitoring information.

  For example, when the topology of the de Bruijn graph structure is configured by the servers A1 to An, since one server monitors two monitored servers as shown in FIG. When a query from the server management apparatus M is received, this query is transmitted to two monitored servers, when a query is received from another server, this query is transmitted to two monitored servers, and two other A function that receives a query from the server and returns a second notification to the query source, and a response from the two monitored servers to the first query source. A function for returning the monitoring information stored in the monitoring information storage means 13 and, if the reply from the monitoring target server is monitoring information, the monitoring information is stored in the monitoring information storage means 13. And a function of adding the stored monitored information. With such a configuration, the monitoring information inquiry reply unit 15 of the server that first starts sending the monitoring information receives the reply that the inquiry is the second time from both of the two monitoring target servers, and receives the monitoring information. Starts sending.

  By the monitoring information inquiry reply means 15 having such a function, an inquiry transmitted from the server management apparatus M to one server A1 relays and propagates each server along a monitoring topology having a specific graph structure, Queries are distributed to all servers A1 to An. Each server adds the monitoring information received from the monitoring target server to the monitoring information stored in its own monitoring information storage means 13 and sends it back to the inquiry source. The received server A1 holds monitoring information related to all the servers A1 to An in which monitoring information from each server is accumulated, and the server management apparatus M acquires the monitoring information related to all the servers A1 to An from this server A1. be able to.

  Here, the monitoring means 12, the monitored means 14, the monitoring information inquiry reply means 15, and the monitoring topology information update means 16 in the present embodiment may be configured such that their function contents are programmed and executed by a computer. .

  As described above, the servers A <b> 1 to An in the present embodiment include the monitoring unit 12 and the monitored unit 14, thereby having two roles of a monitoring side that monitors other servers and a monitored side that is monitored. Also, the servers A1 to An are provided with the monitoring information inquiry reply means 15, so that when the monitoring information inquiry is received from the server management apparatus M, the inquiry is propagated to all the servers A1 to An along the monitoring topology and returned. The monitoring information of all the servers A1 to An is returned to the inquiry source.

  Next, the monitoring information collecting means 21 in the server monitoring apparatus M is configured so that each server is based on performance information (monitoring information) related to all recently collected servers A1 to An continuously at a constant time interval or a random time interval. The server having the least load is selected, a query for monitoring information is transmitted to the selected server, and the monitoring information relating to all the servers A1 to An is acquired from this server.

  Thereby, since the server monitoring apparatus M communicates with the selected one or a small number of servers and acquires monitoring information (performance information) related to all the servers A1 to An, compared with a method of communicating with each server individually. The load on communication is reduced, and the performance information related to the servers A1 to An can be collected at an appropriate cycle. In addition, since the server that performs communication is selected according to the load situation, the communication load is not concentrated on a specific server.

  The monitoring information management means 22 has a function of storing monitoring information related to all the servers A1 to An and displaying it on a screen of a display unit (not shown). An operation manager such as a data center refers to the performance information of each of the servers A1 to An displayed on the screen and manages the operations of the plurality of servers A1 to An.

  When receiving the identification information from the server newly connected to the network L, the monitoring topology configuration unit 24 reconfigures the monitoring topology having a specified graph structure by a plurality of servers including the new server, and adds the new topology. It has a function of transmitting connection destination server information to the server and transmitting the changed connection destination server information and monitoring ID to the server whose monitoring ID and connection destination server have been changed by this reconfiguration.

  Here, the monitoring information collection unit 21, the monitoring information management unit 22, the abnormality detection unit 23, and the monitoring logical monitoring topology configuration unit 24 in the present embodiment are configured such that the function content is programmed and executed by a computer. May be.

  As described above, the server management apparatus M according to the present embodiment selects a server with the least load, transmits a monitoring information inquiry to the server, and receives monitoring information relating to all the servers A1 to An from the server. 21 is provided, the communication load is reduced as compared with the case where the monitoring information is acquired by directly communicating with each of the plurality of servers A1 to An. In addition, since a server that communicates by sending a query of monitoring information is selected based on the load status each time, the load of communication processing is not concentrated on a specific server.

  Next, the overall operation of the server monitoring system of this embodiment will be described. Here, the following operation description is an embodiment of the server monitoring method of the present invention.

  FIG. 4 is a sequence diagram illustrating a procedure in which the server management apparatus M registers a server newly connected to the network L as a management target. Here, a server newly connected to the network L is assumed to be a server A1.

  First, when the new server A1 is connected to the network L, the identification information notifying means (not shown) in the server A1 transmits server identification information such as its own IP address to the server management apparatus M (FIG. 4). Step S1).

  The monitoring topology configuration unit 25 in the server management apparatus M receives and registers server identification information from the new server A1. Subsequently, the monitoring topology configuration unit 24 assigns a monitoring ID to each server so that all the servers A1 to An including the server A1 form a monitoring topology having a prescribed graph structure (step S2 in FIG. 4). .

  Subsequently, the monitoring topology configuration unit 24 in the server management apparatus M transmits the monitoring ID newly assigned to the server A1 and the information of the server that is the connection destination of the server A1 (step S3 in FIG. 4). At this time, the monitoring topology configuration unit 24 in the server management apparatus M newly monitors the monitoring ID and the server to which the connection destination server is changed by the reconfiguration of the topology by newly participating in the monitoring topology of the server A1. And the monitoring ID and the server information of the connection destination after the change are transmitted.

  The monitoring topology information update unit 16 in the server A1 registers the information of the connection destination server received from the server management apparatus M in the monitoring topology information storage unit 11 as the information of the monitoring target server that the own device monitors (step in FIG. 4). S4).

  Next, FIG. 5 is a sequence diagram illustrating an example of operations related to server monitoring and abnormality detection. Here, it is assumed that the server on the monitoring side is the server A1, and the server on the monitored side to be monitored is the server A2.

  First, the monitoring unit 12 of the monitoring server A1 sends a performance information (operation status) transmission request to the monitoring target server A2 based on the address information stored in the monitoring topology information storage unit 11 at a certain time interval or Transmission is repeated at random time intervals (step S6 in FIG. 5). The performance information is numerical information indicating the operating status of the server, and includes, for example, a CPU usage rate, a memory consumption amount, an HDD free capacity, and the like.

  Then, the monitored means 14 of the server A2 receives the performance information transmission request from the server A1, and returns the performance information of the own server to the server A1 (step S7 in FIG. 5). The monitoring unit 12 of the monitoring server A1 receives the performance information from the server A2, and stores this performance information in the monitoring information storage unit 13 (step S8 in FIG. 5).

  When an abnormality occurs in the server A2, even if a monitoring information transmission request is transmitted from the server A1 side to the server A2 side (step S9 in FIG. 5), the server A2 side cannot make a reply.

  The monitoring unit 12 of the server A1 determines that an abnormality has occurred in the server A1 when communication with the server A2 fails or when there is no reply from the server A2 even after a predetermined time has elapsed (FIG. 5). Step S10). Then, the monitoring unit 12 of the server A1 transmits a report indicating that an abnormality has occurred in the server A2 to the server management apparatus M (step S11 in FIG. 5).

  The abnormality detection unit 23 of the server management apparatus M receives the abnormality report from the server A1 side, and the monitoring logical monitoring topology configuration unit 24 reconfigures the monitoring topology excluding the server A2 where the abnormality has occurred (FIG. 5). Step S12), the changed monitoring ID and server information to be monitored are transmitted to the server in which the monitoring ID and the server to be monitored have been changed by reconfiguration.

  Next, FIG. 6 is a sequence diagram showing an operation of collecting the monitoring information accumulated in the monitoring information accumulation means 13 of the servers A1 to An. The operation illustrated in FIG. 6 is described for the case where a monitoring topology having a de Bruijn graph structure is configured by a plurality of servers A1 to An.

  First, the monitoring information collection means 21 of the server monitoring device M determines the load status of each server A1 to An based on the monitoring information (performance information) related to all the servers A1 to An collected recently, and determines the server with the least load. The selected server is determined as an initial inquiry destination for the inquiry of monitoring information first (step S13 in FIG. 6). Here, it is assumed that the initial inquiry destination is determined as the server A1.

  Then, the monitoring information collecting unit 21 of the server monitoring device M transmits a monitoring information inquiry to the server A1 that is the initial inquiry destination (step S14 in FIG. 6).

  The monitoring information inquiry reply means 15 of the server A1 receives the monitoring information inquiry from the server monitoring apparatus M, and transmits the inquiry to the two monitoring target servers set based on the monitoring topology of the de Bruijn graph structure (FIG. 6). Step S15).

  The server that receives this inquiry transfers the inquiry to the corresponding monitoring target server, and distributes the inquiry to all the servers A1 to An along the monitoring topology. Since the monitoring topology has a de Bruijn graph structure, all the servers A1 to An receive the monitoring information query twice.

  Here, when the monitoring information inquiry reply unit 15 of the server An receives the first inquiry (Q1) (step S16 in FIG. 6), the inquiry is transferred to the two servers to be monitored (step S17 in FIG. 6).

  When the monitoring information inquiry reply unit 15 of the server An receives replies from the two monitoring target servers (notification that both are the second inquiry) (steps S18 and S19 in FIG. 6), the monitoring information storage is performed. The monitoring information in the means 13 is transmitted to the transmission source of the first inquiry (Q1) (step S20 in FIG. 6). When the second inquiry (Q2) is received (step S21 in FIG. 6), a notification to the second time is returned to the inquiry transmission source (step S22 in FIG. 6).

  The monitoring information transmitted from the server An relays the monitoring information inquiry reply means 15 of each server along the reverse direction of the monitoring topology composed of the servers A1 to An, and is held in each server every time it is relayed. Monitoring information is cumulatively added.

  When the monitoring information inquiry reply unit 15 of the server A1 receives the monitoring information from the monitoring target server (step S23 in FIG. 6), it adds it to the monitoring information stored in the monitoring information storage unit 13 of the own device (step S24 in FIG. 6). ).

  When the monitoring information inquiry reply unit 15 of the server A1 receives the monitoring information from the other monitoring target server (step S25 in FIG. 6), the monitoring information stored in the monitoring information storage unit 13 of the own device is received. In addition, the monitoring information relating to all servers is stored in the monitoring information storage means 13 (step S26 in FIG. 6), and the monitoring information stored in the monitoring information storage means 13 is transmitted to the server management apparatus M ( Step S27 in FIG.

  Then, the monitoring information collection unit 21 of the server management apparatus M receives the monitoring information related to all the servers A1 to An from the server A1, and the monitoring information management unit 22 stores the monitoring information (step S25 in FIG. 6).

  As described above, according to the server monitoring system of this embodiment, the monitoring unit 12 of each server A1 to An monitors other servers, so that the server monitoring process is distributed, and the servers A1 to An are enormous. Even if it exists, the monitoring with respect to each server can be performed with a suitable period.

  In addition, by configuring a monitoring topology that logically connects all the servers A1 to An, if a query is sent to any server, the query can be propagated to all the servers. Even if .about.An is an enormous number, monitoring information relating to all servers can be obtained from any server only by sending a query for monitoring information to one or a few arbitrary servers. Therefore, it is possible to collectively collect performance information about all the servers A1 to An at an appropriate cycle without reducing the load applied to the communication of the server management apparatus M and constantly applying a load to a specific server.

  The present invention can be applied to uses such as server monitoring in a data center having more than several thousand servers.

It is a block diagram which shows the structure of the server monitoring system of one Embodiment concerning this invention. It is a functional block diagram which shows the internal structure of the server and server management apparatus in embodiment disclosed in FIG. It is explanatory drawing which shows the specific example of a de Bruijn graph structure. It is a sequence diagram which shows an example of the whole operation | movement of the server monitoring system disclosed in FIG. It is a sequence diagram which shows the other example of the whole operation | movement of the server monitoring system disclosed in FIG. It is a sequence diagram which shows the other example of the whole operation | movement of the server monitoring system disclosed in FIG.

Explanation of symbols

A server M server management device L LAN
e Server internal information 11 Monitoring topology information storage means 12 Monitoring means 13 Monitoring information storage means 14 Monitored means 15 Monitoring information inquiry reply means 16 Monitoring topology information update means 21 Monitoring information collection means 22 Monitoring information management means 23 Abnormality detection means 24 Monitoring topology configuration means

Claims (9)

In a server monitoring system in which a plurality of servers and server management devices are connected to each other via a line,
Each of the servers
The server management device is designated as the connection destination of the own server so that the plurality of servers constitute a monitoring topology having a Doubling-Yen graph structure or a Kauts graph structure, or information related to other predetermined servers is stored. Monitoring topology information storage means;
Monitoring means for acquiring another server related to the stored information as a monitoring target server, and repeatedly acquiring performance information of the monitoring target server at a preset time interval;
Monitoring information storage means for storing the acquired performance information of the monitoring target server as monitoring information;
A monitoring information inquiry transmitted from the server management device is received and transferred to the monitoring target server. When a query transferred from another server is received, the inquiry is relayed to the monitoring target server. When the monitoring information stored in the monitoring information storage means is returned to the inquiry from the server management apparatus or the other server and the monitoring information returned from the monitoring target server is received. Monitoring information inquiry reply means for relaying the monitoring information and transmitting to the transmission source of the inquiry,
The server management device
Based on the monitoring information on all servers recently collected at a preset time interval, the server with the lowest load is sequentially selected from the plurality of servers, and an inquiry for monitoring information is sent to the selected server. Monitoring information collecting means for receiving monitoring information related to all servers returned to the server;
With monitoring information management means for managing the monitoring information related to all the received servers,
The monitoring information inquiry return means in each server is the second and subsequent queries instead of the monitoring information to the transmission source of the query received after the second time when an equivalent query is received from different transmission sources. A function to send back notifications to the effect
Server monitoring system characterized by comprising a. In the server monitoring system according to claim 1 ,
The monitoring information inquiry reply means in each server confirms the reply from the monitoring target server that transferred the inquiry when returning the monitoring information stored in the monitoring information storage means to the inquiry transmission source. When the response from the monitoring target server is the monitoring information, the monitoring information is added to the monitoring information stored in the monitoring information storage means and transmitted to the inquiry transmission source. A server monitoring system characterized by that. In the server monitoring system according to claim 1 ,
The monitoring means in each server determines whether an abnormality has occurred in the monitoring target server based on the performance information acquisition result of the monitoring target server, and determines that an abnormality has occurred in the monitoring target server A function to notify the server management device that an abnormality has occurred in the monitored server
The server management device
When a notification notifying the abnormality of the corresponding monitored server is received from each of the servers, the monitoring topology is reconfigured on a plurality of servers other than the server on which the abnormality has occurred, and the reconfigured server is changed to a connection destination server. A server monitoring system comprising monitoring topology configuration means for transmitting information of a connection destination server after change to a server having a problem. In the server monitoring system according to claim 3 ,
When the monitoring topology configuration means in the server management device receives identification information transmitted from a server connected to a new line, the monitoring topology configuration unit reconfigures the monitoring topology with a plurality of servers including the new server, and A server having a function of transmitting information of a connection destination server corresponding to a new server, and transmitting information of the connection destination server after the change to a server whose connection destination server has been changed by this reconfiguration Monitoring system. A server in a server monitoring system in which a plurality of servers and a server management device are connected to each other via a line,
The server management device is designated as the connection destination of the own server so that the plurality of servers constitute a monitoring topology having a Doubling-Yen graph structure or a Kauts graph structure, or information related to other predetermined servers is stored. Monitoring topology information storage means;
Monitoring means for acquiring another server related to the stored information as a monitoring target server, and repeatedly acquiring performance information of the monitoring target server at a preset time interval;
Monitoring information storage means for storing the acquired performance information of the monitoring target server as monitoring information;
A monitoring information inquiry transmitted from the server management device is received and transferred to the monitoring target server. When a query transferred from another server is received, the inquiry is relayed to the monitoring target server. When the monitoring information stored in the monitoring information storage means is returned to the inquiry from the server management apparatus or the other server and the monitoring information returned from the monitoring target server is received. Monitoring information inquiry reply means for relaying the monitoring information and transmitting to the transmission source of the inquiry,
When the monitoring information inquiry reply means receives an equivalent inquiry from different transmission sources, the monitoring information inquiry reply means informs the transmission source of the inquiry received after the second time that the inquiry is the second time or later instead of the monitoring information. Ability to reply notifications
Servers comprising the. The server according to claim 5 , wherein:
The monitoring information inquiry reply means, when returning the monitoring information stored in the monitoring information storage means to the inquiry transmission source, confirms the reply from the monitoring target server to which the inquiry has been transferred, and returns the reply. When the reply from the monitoring target server is the monitoring information, the monitoring information is added to the monitoring information stored in the monitoring information storage means and transmitted to the transmission source of the inquiry. A server characterized by In a server monitoring system in which a plurality of servers and server management devices are connected to each other via a line,
Each of the servers is designated as the server management device as a connection destination of its own server so that a plurality of servers form a monitoring topology of a Doubling graph structure , or another predetermined server is set as a monitoring target server The performance information of the monitoring target server is repeatedly acquired at a preset time interval, and the acquired performance information is stored as monitoring information in a monitoring information storage unit provided in advance,
The server management device sequentially selects the server with the lightest load from the plurality of servers at a preset time interval based on recently collected monitoring information about all servers, and queries the selected server for monitoring information Send
The server that receives the query from the server management device transmits the query to two preset monitoring target servers, and relays this query to all servers by relaying the query along the monitoring topology.
Each server relays this inquiry of monitoring information along the monitoring topology and transmits it to all servers,
The server that has received the inquiry from the server monitoring device or another server returns the monitoring information stored in the monitoring information storage means to the transmission source of the inquiry,
Each server relays and transmits the monitoring information related to each server along the reverse direction of the monitoring topology,
The server management device receives monitoring information from all servers from the server that sent the monitoring information inquiry directly , and
The server that received the inquiry from two other servers returns a notification to the sender of the second inquiry that it was the second time,
A server that has received a reply to the inquiry from both of the two monitoring target servers returns the monitoring information stored in the monitoring information storage means to the transmission source of the first inquiry,
The server that received the monitoring information reply from the monitoring target server adds the monitoring information stored in the monitoring information storage means of the own server to the monitoring information from the monitoring target server, and sends the first inquiry To the monitoring information related to each server, and relaying the transmission information while accumulating each server along the reverse direction of the monitoring topology,
A server monitoring method, comprising: receiving monitoring information related to all servers accumulating monitoring information related to each server from a server to which the server management apparatus directly sends an inquiry for the monitoring information . In each server in a server monitoring system in which a plurality of servers and server management devices are connected to each other via a line,
The server management apparatus is designated as the connection destination of its own server so as to configure a monitoring topology having a Dobrien graph structure with the plurality of servers, or another predetermined server is set as the monitoring target server A monitoring function for repeatedly acquiring the performance information at a preset time interval and storing the acquired performance information in the monitoring information storage means provided in advance as monitoring information;
A monitoring information query is received from the server management device, and this query is transmitted to two preset monitoring target servers. When this query is received from another server , the query is relayed according to the monitoring topology. Monitoring information inquiry function to be sent to the monitoring target server,
In response to an inquiry from the server management apparatus or the other server, the monitoring information stored in the monitoring information storage unit is returned to the inquiry transmission source, and the monitoring information returned from the monitoring target server is received. While causing the computer to execute a monitoring information reply function that relays this monitoring information to the inquiry transmission source ,
When the monitoring information reply function receives an equivalent inquiry from different transmission sources, the notification that informs the transmission source of the inquiry received after the second time that the inquiry is the second time or later instead of the monitoring information A server monitoring program characterized by having a function of returning a message . In the server monitoring program according to claim 8 ,
The monitoring information reply function executes a reply after confirming a reply from the monitoring target server that has transferred the inquiry when returning the monitoring information stored in the monitoring information storage means to the inquiry transmission source. When the reply from the monitoring target server is the monitoring information, the monitoring information is added to the monitoring information accumulated in the monitoring information accumulating means and transmitted to the inquiry transmission source. A server monitoring program characterized by being.

Images