The present application claims priority from Japanese application JP 2005-130735 filed on Apr. 28, 2005, the content of which is hereby incorporated by reference into this application.
The present invention relates to a modular type computer system, and more particularly to the storage module of a blade server.
A rapid growth in the Internet business increases the processing demand for a computer system with an apparent increase in the management cost.
In general, a server system in the Internet business is built based on a three-tier model in which the system is built in three tiers, front-end, mid-tier, and back-end, to meet each processing need.
A blade server, in which multiple servers are installed densely in one chassis, is used for a front-end server that performs the services of a Web server because the dependence among the servers is low. On the other hand, a standalone SMP (Symmetric Multi Processor) server is used for a back-end server that functions primarily as a DB server because the data processing load is heavy and data integrity must be ensured.
Conventionally, different server devices are used for those functional units to build a system. One of the problems with such a system is that the processing performance cannot be allocated among those server devices and, therefore, the devices must be added or replaced individually to meet a change in the application requirements. In a system that is built based on the conventional method described above, TCO (Total Cost of Ownership) cannot be reduced because the hardware is distributed and it is difficult to improve the server operating rate of the whole system.
To solve the problem described above, the focus of attention is now on the integration of the hardware, especially, on the integration of various server functions that is made possible by the expanded application of a blade server. A conventional blade server, which can contain only one or two built-in disks because of its high-density installation feature, has a limited server application. However, if a flexible disk configuration, such as the RAID configuration or JBOD (Just Bunch of Disks), is used to make it possible to provide a server device that expands the application of each blade server to increase the efficiency of the integration of the distributed hardware, a server device useful for server consolidation will be provided.
Although conventional technologies for a modular type computer system such as a blade server are disclosed, for example, in US2003/0090878A1, US2004/0054780A1, and JP-A-2004-355351, a flexible disk configuration for use in a blade server is not yet disclosed.
As described above, a general blade server has one or two built-in disks and, at the RAID level, mirroring is carried out at best. Increasing the number of disks and increasing the flexibility in the disk configuration with the use of RAID5 or spare disks are efficient for building a high-reliability, high-installation-density server.
It is an object of the present invention to provide a blade server that allows disks to be added to a server in one chassis.
The present invention provides a blade server system comprising a plurality of server modules and a management module that manages the whole of the system. The server module slots are connected via the disk interface on the back plane. One or more storage modules, each having a hard disk, are installed in any server modules slot to allow the storage modules to function as additional disks or specific server modules.
According to the present invention, the management module has an interface for identifying the server modules and the storage modules installed in the server slots, the interlock between the server modules and the storage modules is defined by the module identification information, and the power interlock between the server modules and the storage modules is controlled according to the defined interlock.
The blade server system according to the present invention allows an additional disk of a server to be installed on a chassis as a module that is slot-compatible with a server module, thus enabling a server having a disk is added according to the application need and an ordinary server to be mixed in one chassis. Therefore, as compared with a general blade server, the blade server according to the present invention increases the server configuration flexibility and the installation density.
Other objects, features and advantages of the invention will become apparent from the following description of the embodiments of the invention taken in conjunction with the accompanying drawings.
In the server module 11, a memory 22 is connected to a CPU 23, which performs the processing of various applications, via a chip set 21. The chip set 21 is connected to a hard disk 24, installed in a storage module 12, via the SCSI bus on a back plane 15. A module management unit 20 collects various monitor signals of the server module and sends the collected monitor signals to a management module 13 via a I2C I/F. The module management unit 20 receives the power control signal from the management module 13 and instructs a DC-DC regulator 18 to output and shut down the power in the module. The module management unit 20 sends module identification information to the management module 13 via the I2C I/F. Note that a standby power is constantly supplied to all module management units.
The storage module 12, in which one or more hard disks 24 are installed, is connected to the SCSI I/F on the back plane. A module management unit 30 collects the status of the installed hard disks and the monitor signal of the storage module and notifies them to the management module 13. In addition, the module management unit 30 receives the power control signal from the management module 13 to control the output and the shutdown of the power in the module.
The management module 13 uses a service processor 16 to turn on, shut down, and monitor the power of the modules via the I2C I/F. In addition, the management module 13 saves the identification information on the modules, installed in the slots, in a configuration information file 17 as the device configuration information. The management module 13 is inserted into a management module slot (not shown in drawings).
A power module 14 converts the AC power input to the DC power via an AC-DC converter 19 and supplies the power to the modules via the power-supply line on the back plane 15. Two or more power modules 14 may be connected in parallel in a redundant configuration.
The back plane 15 is a board for wiring various I/Fs among modules. Although the block diagram shown in
When modules are installed in the slots, the management module 13 acquires information on the types of modules installed in the slots via the I2C I/F and saves the acquired information in the configuration information file 17 as the device configuration information. To determine whether or not the server module and the storage module in specific slots are connected, the management module holds the I/F connection contents on the back plane as the system-specific back plane connection information and compares the information with the device configuration information in the configuration information file 17. In the example in
When a request to turn on the power of the server module 11 is issued in the above condition (step 33), the service processor 16 of the management module 13 refers to the interlock configuration information in the configuration information file 17 (step 34) and sends the power-on signal to the module management unit 30 of the storage module 12, with which the server module 11 is interlocked, via the I2C I/F. In response to the power-on signal, the module management unit 30 of the storage module 12 turns on the power of the storage module 12 (step 35). If there is no storage module connected to the server module 11, a warning message is output (step 36).
Upon recognizing that the storage module 12 is turned power on, the management module 13 sends the power-on signal to the module management unit 20 of the server module 11 via the I2C I/F. In response to the power-on signal, the module management unit 20 of the server module 11 turns on the power of the server module 11 (step 37). The management module 13 checks the power-on state (step 38) and, if an abnormal condition is found, outputs a warning message (step 39). The above procedure completes the power interlock between the storage module 12 and the server module 11.
The server modules 11 are installed in slots #0 and #1, the storage modules 12 are installed in slots #2 and #3, and the status is saved in device configuration information 50 shown in
When one storage module is installed but there are multiple paths to server modules, the information is written in the interlock configuration information 70 as follows. When the modules are installed in the slots, the system-defined initial connection mode (for example, the mode in which the storage module is connected to a server module with the smallest slot number) is written as the interlock configuration information 70 in the configuration information file 17. To change the connection mode, the user can rewrite the interlock configuration information 70 in the configuration information file 17 via the service processor 16 before turning on the power.
A SCSI controller 26 of the IO module 25 is connected to the chip set 21 of the server module 11 via the IO I/F (PCI bus, etc.) on the back plane 45, and the SCSI controller 26 is connected to the hard disk 24 of the storage module 12 via the SCSI bus on the back plane 45. A module management unit 40 of the IO module 25 collects various monitor signals of the IO module 25 and sends the collected monitor signals to the management module 13 via the I2C I/F. In addition, the module management unit 40 of the IO module 25 receives the power control signal from the management module 13 and instructs the DC-DC regulator 18 to output or shut down the power in the module.
The following snows the power interlock procedure in the embodiment shown in
When the modules are installed in the slots, the management module 13 acquires the configuration of the modules installed in the slots via the I2C I/F. In the example shown in
When a request to turn on the power of the server module 11 is issued in the above condition, the service processor 16 of the management module 13 refers to the configuration information file 17 and sends the power-on signal to the module management units of the IO module 25 and the storage module 12, which are interlocked with the server module 11, via the I2C I/F. In response to the power-on signal, the module management units of the IO module 25 and the storage module 12 turn on the power of the modules.
Upon recognizing that the IO module 25 and the storage module 12 are turned power on, the management module 13 sends the power-on signal to the module management unit 20 of the server module 11 via the I2C I/F. In response to the power-on signal, the module management unit 20 of the server module 11 turns on the power of the server module 11. The above procedure completes the power interlock between the storage module 12 and the server module 11.
As described above, the blade server system according to the present invention allows a storage module to be installed in a server module slot, thus enabling the disks of the blade server to be configured flexibly.
It should be further understood by those skilled in the art that although the foregoing description has been made on embodiments of the invention, the invention is not limited thereto and various changes and modifications may be made without departing from the spirit of the invention and the scope of the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
2005-130735 | Apr 2005 | JP | national |