Electrical computers and digital processing systems: memory – Storage accessing and control – Control technique
Reexamination Certificate
1999-08-19
2004-01-20
Ellis, Kevin L. (Department: 2188)
Electrical computers and digital processing systems: memory
Storage accessing and control
Control technique
Reexamination Certificate
active
06681303
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a technique of copying of date between storage systems without the intervention of a CPU and a technique of arrangement/rearrangement of a logical volume on RAID groups in a storage system. Further, the present invention relates to a storage system in an information processing system or the like and relates to a computer system having a function of generating a copy of data stored in a volume.
A technique of remote copying exists as one of techniques in which the copying of data is performed between storage systems.
In the remote coping, the writing of data in a duplicated manner is made without the intervention of a CPU between a plurality of storage systems located at physically remote places. Therein, storage systems respectively arranged at primary and secondary sites are connected by a dedicated line or public line. Also, a logical volume having the same capacity as that of a logical volume existing on the storage system of the primary site and made an object subjected to copying (hereinafter referred to as a copy source logical volume) is formed on the storage system of the secondary site as a logical volume which is paired with the copy source logical volume (and will hereinafter be referred to as a copy destination logical volume). Then, data of the copy source logical volume of the primary site is copied into the copy destination logical volume. Also, in the case where the updating of data of the copy source logical volume of the primary site is made from a CPU, the updated data is transferred to the storage system of the secondary site and is then written into the copy destination logical volume. Thus, in the technique of remote copying, the duplicated state of a logical volume is always held at the primary and secondary sites.
Therefore, even if the primary site becomes impossible of use due to natural disasters such as earth-quake and flood or artificial disasters such as fire and terrorism, it is possible to rapidly restart the service by use of the logical volume on the storage system of the secondary site.
A known prior art relevant to the remote copying includes a technique disclosed by U.S. Pat. No. 5,155,845. The known technique of performing the copying of data between storage systems also includes a technique of migratory copying (or data migrating copy) disclosed by U.S. Pat. No. 5,680,640.
According to the known migratory copying technique, in the case where a new storage system is introduced in lieu of a storage system hitherto used by a customer, a processing for copying data of a logical volume on the old storage system into the new storage system is realized in the following manner.
Namely, the destination of connection of a CPU is changed from the old storage system to the new storage system. Further, the new storage system and the old storage system are connected. While receiving an input/output request from the CPU, the new storage system reads data from a logical volume on the old storage system and copies the read data into a logical volume on the new storage system (that is, performs a migratory copying).
With this technique, since the copying of data between the logical volumes of the new and old storage systems can be performed without the intervention of the CPU, a load imposed on the CPU at the time of data migration is eliminated, thereby enabling the data migration even when the service is being performed.
In “VA Case for Redundant Arrays of Inexpensive Disks (RAID)”, Proc. ACM SIGMOD, June 1988, G. A. Patterson, G. Gibson and R. H. Katz of the University of California, Berkeley, U.S.A. have given a taxonomy of five organizations of disk arrays as RAID levels to evaluate the storage cost, performance and reliability of each RAID level. The RAID levels result from the classification of a redundant array forming method in the case where a storage system is structured using inexpensive disk devices. Therein, the redundant array forming method is classified in accordance with a data allocating method and a redundant data generating method. RAID's 1, 3 and 5 in the taxonomy of five organizations are presently applied to many products. These RAID levels have the following characteristics.
RAID 1 (Mirrored Disks): The same data is held by different disk devices. Since data is duplicated, the reliability is high but the storage cost is doubled.
RAID 3: Data is divided into units of several bytes so that they are allocated to a plurality of data disk devices. Redundant or check data is generated by an exclusive OR of divisional data and is stored in another or one redundant disk. Since all the disk devices synchronously operate for the input/output of data, an excellent performance is exhibited in the case where the input/output of long or large data is performed. On the other side, the RAID 3 is unsuitable for an on-line transaction processing or the like in which short data is randomly accessed.
RAID 5: Data is divided into units of blocks and the data blocks are distributively allocated to a plurality of disk devices. Redundant data is generated by an exclusive OR of divisional data and is stored at predetermined positions on storage devices. In the RAID 5, respective redundant blocks are distributively allocated to the disk devices so that all the disk devices include the redundant blocks. Thereby, a load imposed on the disk device at the time of access to redundant block is distributed. When the data block is updated, a disk access is generated in order to recalculate the corresponding redundant block, thereby deteriorating the performance. This is called write penalty.
The RAID 5 is characterized in that if the size of data to be accessed does not exceed the size of the block, the access to only one disk device suffices and hence the plurality of disk devices can operate independently, unlike the RAID 3. Therefore, the RAID 5 is suitable for an on-line transaction processing in which relatively small data is randomly accessed.
As mentioned above, the characteristics in the aspects of reliability, cost and performance are provided in accordance with each RAID level. In actual services, it is preferable that the optimum RAID level is selected taking those characteristics into consideration and in accordance with the property of the service.
An assembly of storage devices realizing a certain RAID level or an assembly of partial areas of storage devices is called a RAID group, and one RAID level is realized by this RAID group. A logical volume which a CPU makes an object of input/output is generally mapped on one RAID group by virtue of storage devices.
Also, there exists a technique of acquiring the backup of consistent data without stopping the updating for a volume.
It is generally known that the backup is acquired as means for preventing important data from being fully lost when a fault is generated in a storage device. It is general that in order to assure the consistency of data subjected to the acquisition of backup, a write/read processing for the corresponding volume is stopped during a time when the backup is being acquired. Accordingly, there is a problem that during the time when the backup is being acquired, a processing must be stopped which uses a volume made an object of backup. According to a known method of solving this problem, a copy of a volume is generated in a storage device so that (1) normally, data of the original volume and data of the copy volume are made coincident with each other, (2) during a time when the backup is acquired, the data of the original volume and the data of the copy volume are not made coincident (and hence the copy volume represents the original volume at a certain point of time when the consistency is assumed), and (3) the copy volume is used for the backup. Thereby, it is possible to acquire consistent data as backup data without stopping the processing during the time when the backup is acquired.
SUMMARY OF THE INVENTION
In the conventional technique of remote copying, since the unit of an object of copying is a logical volume, as
Arai Kouji
Nakamura Katsunori
Oeda Takashi
Watanabe Haruaki
Yamagami Kenji
Ellis Kevin L.
Hitachi , Ltd.
Mattingly Stanger & Malur, P.C.
LandOfFree
Storage 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 system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Storage system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3196468