JP2009171476A - Server management system and its construction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that the necessity for a large amount of communication occurs, since it is necessary for a management server to communicate with all servers being managing objects and monitor their states, to confirm whether all operating servers are operated normally on the occasion of managing a large number of servers. <P>SOLUTION: With respect to a system wherein communication is performed between a management server and a large number of servers being managing objects, a highest rank representative server is determined from among the servers of managing objects, and low order servers are formed into a group. Moreover from among the formed group, a representative server is determined, and moreover a low-order group is formed. By acquiring a mechanism for repeating these procedures, a hierarchical structure is constructed automatically. In this system, only the highest order representative server communicates with the management server, by using this constructed hierarchical structure, while putting together the management information of the managing object servers instead of making all the servers communicate with the management server. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a communication method used when a management server manages a plurality of servers in an environment in which the plurality of servers are operating.

  When a system abnormality is detected and reported, or when a system is informed regularly, a maintenance system is widely adopted in which monitoring is performed by a managed server reporting to a receiving center. In recent years when networks such as the Internet and communication technologies have advanced, a maintenance system that uses the Internet network and reports using IP (Internet Protocol) is common. The features of this maintenance system include the number of servers that can be managed at the receiving center and the number of servers that can be processed simultaneously. In one report receiving center, the number of servers that can be managed is limited, but the number of servers that can be managed is much larger than the number of servers that can be processed simultaneously. The reason is that it is designed considering probabilistically the case where a plurality of reporting devices simultaneously report.

  Prior art relating to monitoring systems is disclosed or proposed in a number of technical literature. There has been disclosed a maintenance system that improves the wiring efficiency between a plurality of monitoring devices and a plurality of devices to be monitored, and improves the efficiency of communication of a single monitoring center with respect to the plurality of monitoring devices (see Patent Document 1). Also disclosed is a method for improving the efficiency of notification when a center that receives a notification monitors each monitoring target and stratifies the monitoring target according to the instructions of the center so that a plurality of monitoring targets can be notified simultaneously. (For example, refer to Patent Document 2). And, in order to avoid the need to report from each server to the report receiving center, a method is adopted in which a device for receiving and reporting a plurality of reports is placed to improve the monitoring efficiency (for example, see Non-Patent Document 1). is there.

JP 2001-23064 A JP 2006-310947 A LogStare Enterprise [terilogy] (http://www.secuavail.com/service/logstareoutline.html)

  When managing a large number of servers, it is necessary to monitor the status by communicating with the management server and all the servers to be managed in order to check whether all the operating servers are operating normally. . Conventionally, in a server management system in which a dedicated line cannot be used between a server to be managed at a customer site and a monitoring center, and a general Internet line must be used, between each server at the customer site and the monitoring center A large amount of communication occurs because of communication. There is an existing means to solve this problem by aggregating many communications. However, when a method for improving the efficiency of monitoring by placing a device that receives a plurality of notifications and reporting (Patent Document 1 or Non-Patent Document 1), it is necessary to place the device in the customer's network, The cost of purchasing a device that is not related to business, the space for arranging the device, the trouble of setting the device, and the failure monitoring of the device itself that consolidates communication are also required. In the case of a system that has a hierarchical structure and the representative monitoring target reports (Patent Document 2), the reporting means when a failure occurs in the representative monitoring target is lost. In addition, when there are many servers to be managed, it takes time to construct a hierarchical structure.

  The present invention solves the above problems.

In the present invention, the system that communicates between the management server and the managed server does not communicate with all the servers, but collects the status of the entire managed server and communicates from the representative server. A method of providing a standby server as a countermeasure when a server failure occurs. All servers have the server information table of this system, and the construction of this system is realized by executing the following procedure.
(1) The highest representative server that performs direct communication from the server group to the management server is determined.
(2) The highest representative server is defined as a representative server.
(3) All the servers that can be managed by the representative server are assigned as a group to the lower level.
(4) A standby server is assigned from the lower group assigned in (3).
(5) One of the subordinate groups assigned in (3) is set as a representative server.
(6) A group of only a manageable number of servers is assigned to the representative server assigned in (5) above.
(7) A server that is not a standby server and has no lower group assignment is defined as a representative server.
(8) By repeating the steps (3) to (7), all servers are assigned to the group until there are no servers belonging to the group, and the assignment is reflected in the table.
(9) The table created in (8) is expanded to all managed servers.
(10) Each server confirms the connection with the server of the lower group when the representative server of the own server and the lower group are assigned.
(11) The state of the reporting system is mutually monitored between the representative server and the standby server.

  By realizing these mechanisms, a hierarchical structure is automatically constructed.

  The mechanism for automatically constructing the hierarchical structure of the communication system reduces the labor of construction. Furthermore, even if the established hierarchical structure does not communicate with all monitored servers, the highest representative server collects the status of the monitored servers using the hierarchical structure, so that only the highest representative server can By performing communication, it is possible to collect the status of all monitored servers. By adopting such a mechanism, the load on the management server and the route to it is reduced. The degree of reduction of the load increases as the number of monitoring targets increases. Further, since all the monitoring target servers are further provided with standby servers, the communication system is automatically restored even when a failure occurs in the representative server. Further, since only the highest representative server communicates with the management server, it is effective in terms of security.

  Necessary figures will be described in describing the embodiment. FIG. 1 is a diagram showing a configuration example of a communication system targeted by the present invention. FIG. 2 is a diagram showing an internal mechanism of a server necessary for realizing this communication system. FIG. 3 is a diagram showing communication system basic data in FIG. FIG. 4 is a sequence diagram showing a configuration procedure of the communication system. FIG. 5 is a diagram illustrating an example of information transmitted from a server of a group that is not the highest group in FIG. 1 to a representative server of the upper group. FIG. 6 is a diagram showing an example of information reported from the representative server of the group (G02) to the upper representative server in FIG.

  The description of the examples begins. In this embodiment, it is assumed that the user company uses the communication system shown in FIG. This is a system in which the management server (20) performs alive monitoring of all servers in the site A (10) via the Internet network (30).

  First, the configuration of each server in this system will be described with reference to FIG. The server constituting this system includes communication system basic data (S12), a heartbeat register (S13), and a heartbeat monitoring mechanism (S24).

  Next, a configuration procedure of the communication system will be described. FIG. 4 is a sequence diagram showing a configuration procedure of the communication system, which will be described along this diagram. In the present embodiment, “operating time” is adopted as an example of the evaluation value in FIG. A system construction request was made to the temporary representative server at the request of the user. In the request, the number of parameters that can be managed (m units, 3 units in this example) is given to the temporary representative server by the user (T01). The temporary representative server that has received the request makes a return request to the server in the base by broadcast and performs detection. For the detected server, the temporary representative server goes to the device ID column (C10), IP address column (C12), and operation time column (C13) of all servers in the communication system basic data (S12) of the temporary representative server. Collect and update information (T02). Further, the server (41) having a long operation time is determined as the highest representative server, and the role column in the group of the server (41) in the communication system basic data (S12) is updated to “representative”. The communication system basic data (S12) currently possessed by the temporary representative server is transferred to the highest representative server (41) and the manageable number (m units) is taken over. Now, the representative server is defined as the highest representative server (T03). The highest representative server (41) assigns unassigned servers to the lower group of the representative server in the order of operation time, as many as the number that can be managed by the representative server (m), and the communication system of the highest representative server (41) The group ID column (C10) of the data (S12) is updated (T04). From the group assigned in T04, the server with the longest operation time is determined as the standby server, and the role column (C16) in the group of the communication system basic data (S12) of the highest representative server (41) is updated. (T05). The highest representative server (41) determines whether there is an unassigned server from the group ID column (C10) of the communication system basic data (S12) of its own server (T06). If there is an unassigned server, the representative server is defined as the server with the longest operation time that is not the standby server among the servers to which no group is assigned (T07). By repeating the above steps T04 to T07, all servers are assigned to the group. When the assignment is completed, the highest representative server (41) notifies the assignment to all the servers and finishes (T09). The representative server of each group constitutes a heartbeat with the standby server of the lower group. Thus, the communication system of the present embodiment is constructed.

  Furthermore, the case where a server is added to the communication system of a present Example is demonstrated. Specifically, assume that one server is added in the current configuration (FIG. 1). The highest-level representative server (41) periodically transmits a broadcast for server addition detection. Now, the top representative server (41) has detected the added server. A row is added to the bottom of the communication means table (S11) of the highest representative server, and the new IP address `` 11.203.22.11 '' and the device ID column (C10) are added to the IP address column (C12) of the added row. ) Is collected and updated with the new device ID “74”. If there is a group that is lower than the number of representative servers that can be managed among the lowest-level groups, the highest-level representative server (41) uses the same group ID and hierarchy depth as the group's own communication system basic data. Update the group ID column (C10) and hierarchy depth column (C15) of the added server (74) in (S12). If there is not, the group ID column of the server (74) to which the communication system basic data (S12) of the local server has been added and assigned to the top layer server, even among servers without a lower group assignment (C10) and the hierarchy depth column (C15) and the upper representative server column (C17), even if there is no new group ID and lower group assignment, the depth of the hierarchy one level higher than the hierarchy of the top layer server. And the device ID of the uppermost server is entered even if the upper representative server has no lower group assignment. Thereafter, the highest representative server expands the communication system basic data (S12) of the own server to all monitoring target servers. With the above procedure, a server is added to the communication system of the present embodiment.

  Next, the life and death monitoring method will be described. This will be described with reference to FIG. 3 which is the details of the communication system basic data (S12). The top-level representative server (41) received a request for alive monitoring from the management server (20). The highest representative server (41) makes a life / death monitoring request to the lower representative server (53). Hereinafter, the representative server recursively sends a life / death monitoring request to a lower representative server. Life monitoring is performed from the representative server (63) of the lowest group. Specifically, the representative server (63) of the upper level group (G02) sends a request to the server group of G03, and the server group of G03 is the representative server of G02 (63). ), If the G02 representative server (63) receives a response to the status column (C14) in the communication system basic data (S12) of its own server, the response is confirmed to be “normal”; Update to "failure". When the G02 representative server (63) completes the update of the status column (C14) of the server in the group lower than its own, the communication information basic data (S12) of the G01 representative server (53) in the group lower than itself is updated. Update the server status column (C14). The representative server recursively updates the status column (C14) of the communication system basic data (S12) of the upper representative server, so that the communication system basic data (S12) of the highest representative server (41) is updated. The status information of all servers is updated in the status column (C14). When the update of the status information of all the servers is confirmed, the highest representative server (41) notifies the management server (20). Life and death monitoring is realized as described above.

  Next, the role of the standby server will be described. Each server performs the same communication with the standby server of its own group when communicating with the representative server. The standby server constructs a heartbeat with the upper representative server. In the heartbeat monitoring, when a failure is detected in the upper representative server, the standby server is replaced with the upper representative server. This will be specifically explained with reference to FIG. A failure occurred in the G01 representative server (53), and the standby server (62) detected it. The standby server (62) notifies the highest representative server (41) of the failure. Upon receiving the notification, the highest-level representative server (41) changes the status column (C14) of the server (53) where the failure occurred to the communication system basic data (S12) of its own server from "Normal" to "Fault". To do. Instead of the server (53) that was the G01 representative server, the top-level representative server (41) assigns the G02 standby server (62) as a new G01 representative server, and the communication system basic data of its own server The group role column (C16) of the server (62) in (S12) is rewritten from “standby” to “representative”. The highest representative server (41) further assigns the server (61), which was a normal server of G02, to the G02 standby server, and the role column (C16) of the server (61) of the communication system basic data (S12) of the local server. ) Is changed from “Normal” to “Standby”. Then, the highest representative server (41) expands the communication system basic data (S12) of the own server to all the monitoring target servers. The new G02 standby server (61) and the new G01 representative server (62) that received it construct a heartbeat. From the above mechanism, the communication system can be maintained even when the representative server is in failure.

  If the management server directly monitors the life and death of all servers without using this life and death monitoring system, the number of communications between the management server and the monitoring target server becomes enormous. A great load is applied. If this life and death monitoring system is used, the hierarchical structure is constructed without involvement of the management server, and therefore the communication to the management server is notified collectively by the highest-level representative server. It can be realized with a load.

  In addition, it demonstrates that the communication system structure of a present Example is applicable also about notification other than life and death monitoring. In this embodiment, it is assumed that the user company uses the communication system shown in FIG. This is a system for notifying all reports in the site A (10) via the Internet network (30) to the management server (20) by e-mail.

  The procedure for reporting each server to the management server will be described below. Specifically, an example will be described first with reference to FIG. A reporting factor occurred on server (63). The server (63) refers to the entire basic data (S12) of its own reporting system, and sends the destination “IP address of the upper representative server (I01)”, the source device ID “own server device ID (I02)”, The transmission of data consisting of “the depth of the own server hierarchy (I03)”, the number of notifications “1 (I04)”, the device ID “device ID of the own server (I05)”, and the notification factor “notification factor (I06)” Create and send data (Figure 5). This is received by the representative server (53) of group G1. Furthermore, it is assumed that the representative server (53) has received a report from the server (61). At that time, the representative server (53) performs notification merging and reports to a higher hierarchy. In this case, the destination (I11) “IP address of the host server (41)”, source device ID (I12) “local server device ID (53)”, hierarchy depth (I13) “local server “Depth of hierarchy”, number of reports (I14) “number of reports from server (61) (I24) + number of reports from server (63) (I04)”, and server (61) device ID ( Transmission data (FIG. 6) consisting of I22), notification factor (I26), server ID (I07) and notification factor (I08) of the server (63) is created and transmitted to the upper representative server (40). Then, the top-level representative server (41) that has received the information (FIG. 6) reports to the management server by various known reporting means (E-mail, FTP, SNMP, etc.). In this embodiment, notification is made by e-mail. As a result, in the system that reports directly to the management server (20), two notifications are sent to the management server (20), but in this system, information corresponding to two notifications can be sent with one notification. . In this example, the explanation was made with two notifications for the sake of clarity, but it will be clear that even if there are more than two notifications, it is possible to report with one e-mail in the same sequence.

  If all the servers report directly to the management server via E-mail without using this notification system, the number of reports from all servers and the number of E-mail reports issued to the management server are the same. For this reason, a large number of notifications puts a burden on the management server unless the device is narrowed down. If this notification system is used, many notifications are collected and reported to the management server by e-mail, so even if the site is composed of a large number of servers, the management server can receive light loads. Furthermore, even if a failure occurs in any server, the reporting system can be maintained by the redundant configuration of the representative server and the standby server.

The structural example of the communication system made into object by this invention. The figure which showed the internal mechanism of each server which comprises the component of a system, when implement | achieving this communication system. The figure which showed the communication system basic data in FIG. The sequence diagram which showed the structure procedure of this communication system. The figure which showed an example of the information which the server of a certain group (G02) which is not the highest group in FIG. 1 transmits to a high-order representative server. The figure which showed an example of the information which the representative server of a group (G02) reports to a high-order representative server in FIG.

Explanation of symbols

  (10) ... Location A, (20) ... Management server, (30) ... Internet network, (41) ... Top representative server, (51) ... Server, (52) ... Standby server, (53) ... Representative server, (61) ... Server, (62) ... Standby server, (63) ... Representative server, (71) ... Server, (73) ... Server.

Claims (1)

  In a system in which a plurality of servers are connected to a management server via a network such as the Internet and communicate between the server and the management server, a step of determining a representative server and a server managed under the server And a step of recursively performing the two steps as a server representative of the servers in the group, and automatically forming a hierarchical group, and all the servers manage the management A server management system in which a server representing management information of all servers is aggregated and communicated to the management server without directly communicating with the server.

Images