Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
2002-05-31
2004-09-14
Beausoliel, Robert (Department: 2184)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
Reexamination Certificate
active
06792557
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a storage area network that performs remote copying in the condition that a disk system, which is a storage system, is connected to the storage area network.
BACKGROUND OF THE INVENTION
A magnetic disk unit that has high cost performance is generally used as a device for storing data from a computer. A magnetic disk has a mechanism for reading and writing data by means of magnetic heads that are positioned on both surfaces of each magnetic disk of a plurality of magnetic disks of about 2.5 inch or 3.5 inch size.
The processing time of the magnetic disk, because it operates by mechanical action, is about 10 millisecond, which is slow compared to the processing speed of the processor. There are many cases in which the performance of a system overall does not improve because the processor is made faster but the disk is not made faster. There is the disk array as a means for solving this problem. As described on pages 271-291 of
Understanding I/O Subsystems
, First Edition, by W. David Schwaderer and Andrew W. Wilson, Jr., the disk array is a method that improves performance and reliability by allocating to distribute data to a plurality of drives and also storing redundant data too on the drives. In large-scale systems, a required total capacity of all drives is also large and disk arrays are used because both performance and reliability are required.
The method of achieving high reliability using a plurality of disk arrays over a wide area is described in U.S. Pat. No. 5,870,537 while the disk array increase reliability of the system itself In U.S. Pat. No. 5,870,537, two disk controllers are connected with a mainframe-dedicated optical interface (ESCON), one is defined as a primary disk system and the other as a secondary disk system. There are two host computers, one is connected to the primary disk system and secondary disk system, and the other is connected to only the secondary disk system. In remote copy, when a write request is issued from the host computer which is connected to the primary disk system, to the primary disk system, the primary disk system transfers the write request to the secondary disk system via the aforementioned ESCON and the same data is stored in the secondary disk system. By doing this, even if an error occurs in the storage on one side, the processing is continued by the storage on the other side. Further, in U.S. Pat. No. 5,870,537, the operations when an error occurs in the remote copy system are described. It is described that if an error occurs in the primary disk system, the processing is continued by switching the path from the host computer to the secondary disk system, and when the primary disk system recovers from the error, switching is made between the secondary disk system and primary disk system.
The disk array is feasible in high-speed processing and the fiber channel is highly expected as an interface for connecting disk arrays and host computers. The fiber channel is superior in performance and connectivity, which are deficiencies of SCSI (small computer system interface) generally used in the prior art. Especially, in connectivity, while SCSI can be extended only to a connection distance of a few tens of meters, the fiber channel can be extended out to a few kilometers. It also allows a few times as many devices to be connected. Because the fiber channel allows connection of a wide variety of devices and host computers, it is appropriate for a local area network that is used in data communications between host computers, which is also called a storage area network. The fiber channel is standardized, and if devices and host computers comply with these standards, they can be connected to a storage area network. For example, it is possible to connect a plurality of disk arrays and a plurality of host computers, which have fiber channel interfaces.
However, in the case of aforementioned U.S. Pat. No. 5,870,537, because the dedicated interface is used to connect the disk systems, it is not appropriate for remote copy via a storage area network. Also, in U.S. Pat. No. 5,870,537, if an error occurs in the primary disk system or the secondary disk system, the pair for remote copy cannot recover until the system that caused the error recovers. This is because the dedicated interface is used for connecting the disk systems. This is because when the systems that are a pair for remote copy are connected with a dedicated interface, data can be transferred only between the disk systems that are connected with the dedicated interface. In addition, in U.S. Pat. No. 5,870,537, if an error occurs in the primary disk system, the host computer processing is continued by means of that the host computer switches the I/O destination to the secondary disk system. However, it requires switching on the host computer side and creates a problem of an increased I/O overhead. Further, in U.S. Pat. No. 5,870,537, the connection paths between the host computers and the disk systems are different from the connection path between the disk systems for remote copy. Therefore, an overhead which flows the paths increases during remote copying.
Moreover, in U.S. Pat. No. 5,870,537, there is no description on the case where the primary disk system has been recovered. However, from the facts that the primary site and secondary site are separated and that the secondary host computer can access only the secondary disk system, it is supposed that the primary host computer switches the I/O destination to the primary disk system when recovered from an error.
Disclosure of the Invention
To solve the aforementioned problems, remote copy is performed via a storage area network and also when an error occurs in a storage system for remote copy, a standby storage system connected to the storage area network is assigned as a substitute for the storage system in which the error occurred. Also two host computer adapters that control the connection with a host computer are installed in a disk system which is a storage system. When an error occurs in the primary disk system or the secondary disk system, the processing is performed uninterruptedly by changing the ID of one of the two host computer adapters to the device ID of the disk system that caused the error, without changing the host computer. In addition, a primary disk system that retransfers a command, which has been transferred from a host computer, to the storage area network without changing it, and a secondary disk system that receives a command with an ID different from the own device ID are installed.
REFERENCES:
patent: 5177744 (1993-01-01), Cesare et al.
patent: 5212784 (1993-05-01), Sparks
patent: 5363484 (1994-11-01), Desnoyers et al.
patent: 5615329 (1997-03-01), Kern et al.
patent: 5675723 (1997-10-01), Ekrot et al.
patent: 5870537 (1999-02-01), Kern et al.
patent: 6366987 (2002-04-01), Tzelnic et al.
patent: 6385706 (2002-05-01), Ofek et al.
patent: 6442551 (2002-08-01), Ofek
patent: 6449688 (2002-09-01), Peters et al.
Sato Takao
Takamoto Yoshifumi
Uchigiri Tatsumi
Antonelli Terry Stout & Kraus LLP
Beausoliel Robert
Bonzo Bryce P.
Hitachi , Ltd.
LandOfFree
Storage area network system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Storage area network system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Storage area network system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3265491