Electrical computers and digital processing systems: multicomput – Master/slave computer controlling
Reexamination Certificate
1999-06-15
2003-04-01
Maung, Zarni (Department: 2154)
Electrical computers and digital processing systems: multicomput
Master/slave computer controlling
C714S004110, C710S005000
Reexamination Certificate
active
06542924
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a clustering system for a disk array in which a monitoring client including a manager program and a plurality of user clients are connected with the disk array through a server.
2. Description of the Related Art
A disk array generally has a larger capacity than an HDD (hard disk drive) used for an external memory for a personal computer, etc., and records redundant data in a plurality of the HDDs included in the disk array during operation. By recording such redundant data, recorded data is maintained unchanged even in the event of failures of several HDDs in the disk array. In addition, providing redundancy to the power supply and cooling fans of the apparatus allows parts damaged during operation to be exchanged without interrupting a host computer even if a power supply and a fan are damaged.
Such a disk array is commonly connected with a host computer through a standardized cable via an SCSI (Small Computer System Interface) standardized by ANSI (American National Standards Institute). An SCSI command is transmitted to the disk array from the host computer. The disk array can write therein data from the host computer in accordance with the contents of the SCSI command. The host computer can read data stored in the disk array, and can also read the current states of the HDD and fan.
Control software in the disk array provides the user with functions of displaying on the screen of a computer quality judgment information on whether parts of the disk array are operating normally or they require exchange because of trouble, instructing the user to select several data recording schemes for the disk array, called RAID (Redundant Arrays of Inexpensive Disks), displaying on the screen of the host computer sudden trouble of parts and informing the trouble of addresses of recorded electronic mail.
In recent years, many articles have appeared in which a GUI (Graphical User Interface)is attached to improve operability by permitting a user to view in color the state of the disk array. There are various known operating systems for a disk array. Some examples are Windows 3.x, Windows 95, 98, NT (Microsoft company), Netware (Novel company), HP-UX (Hewlet Packard), and Solaris (Sunmicro System company).
In the prior art, the control software for the disk array operates on the server to which the disk array is connected, and a user enjoys disk array control service at the server. However, enterprise users and the like increasingly actuate a plurality of servers each of which is connected with a disk array, and there are users who operates a server after removing their display from the server. Accordingly, it has become dominant that control of a disk array is centralized, so that a user has no need to go to the disk array, and all services of the disk array can be obtained from one client installed at the user's job site.
To this end, the control software for the disk array, instead of a system wherein a program having all functions is loaded in each server, has adopted a system wherein an agent program is disposed on the client's manager program and each server and the agent program of each server issues instructions to the associated disk array according to instructions received from client's manager program and sends a value received from the disk array to the manager program.
For improving the reliability of a server itself under such conditions, the clustering technique has come into the light spot.
A prior art clustering informing technique will be described.
FIG. 1
is a block diagram illustrating a system configuration which is implemented on Windows NT. This system includes two servers
3
,
4
to effect clustering control of a disk array
5
. A manager program
15
, which is a user interface for the disk array
5
, is loaded on a monitor client
13
, and a master server
3
, a slave server
4
, client
13
, and a general user client
16
are connected with each other through a LAN
12
. Manager program
15
displays information on disk array
5
with a GUI for a user, and receives instructions by keyboard and mouse from the user and provides instructions to the disk array
5
. An agent program
7
and an OS
9
are loaded on master server
3
, and an agent program
8
and an OS
10
are loaded on the slave server
4
. Agent programs
7
and
8
are identical, and receive instructions from manager program
15
, issue an SCSI command to the disk array
5
connected with servers
4
,
5
, and send a reply to the instruction to the manager program
15
based on data returned from disk array
5
. Master server
3
and slave server
4
are connected with each other through a heart beat
11
for mutual communication for the OSs
9
,
10
to realize the clustering. Disk array
5
includes controllers
1
,
2
connected with master server
3
and slave server
4
, respectively, to receive the SCSI command from master server
3
and slave server
4
.
FIG. 2
is a block diagram illustrating the configuration of the clustering system when control is switched to the slave server.
In the configuration of
FIG. 2
, which is basic, when the server suffers from trouble owing to any cause, many general clients connected with each other through the LAN suffer inconvenience. This clustering technique has been devised to prevent the difficulties of the configuration of
FIG. 1
, and as illustrated in
FIG. 2
, the two servers appear to be one highly reliable server when viewed from the client.
First, master server
3
in
FIG. 1
is operated as usual. Specifically, master server
3
communicates with the client
16
on the basis of IP (Internet Protocol) addresses and host names, and an application program in the server is operated. The control software of the disk array
5
is also operated on master server
3
. Slave server
4
and master server
3
recognize mutual conditions at a certain time interval through a method called heart beat communication, and slave server
4
is prevented from being viewed from client
16
.
When master server
3
has trouble at a certain time, slave server
4
detects the failure of master server
3
through the heart beat communication and operates in place of master server
3
. That is, slave server
4
can be viewed from the client, and behaves such that it forces client
16
to recognize the IP address and host name of the slave server as being the same as those of the previous master server. Since the same agent program
8
of the disk array control software as that of master server
3
is included in slave server
4
, the agent program operates on slave server
4
even when the control is switched to slave server
4
. Therefore, manager program
15
at client
13
can communicate continuously and reliably with agent program
8
, ensuring system operation without interruption of the service even when the control is switched to slave server
4
. When the trouble at master server
3
has been corrected, if slave server
4
detects that master server
3
has returned to normal through heart beat communication, the control is shifted to master server
3
, and slave server
4
is put into a standby mode. The control of the transition is guaranteed by both OSs
9
,
10
.
When master server
3
has a trouble owing to any cause, agent programs
7
,
8
can not know by themselves which server is operating between them without a particular cluster API (Application Programming Interface) of the OSs
9
,
10
. Since actually manager program
15
recognize that the system includes one server
25
with higher reliability as shown in
FIG. 2
, it can not know when the control is switched to slave server
4
and when the control is returned to master slave
3
.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a disk array clustering system which is capable of notifying the transition of a server.
To achieve the aforementioned object, a disk array clustering system according to the present invention includes a server transition judgment section.
The se
Chang Jung-won
Foley & Lardner
Maung Zarni
NEC Corporation
LandOfFree
Disk array clustering system with a server transition... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Disk array clustering system with a server transition..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Disk array clustering system with a server transition... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3049708