Electrical computers and digital processing systems: memory – Storage accessing and control – Control technique
Reexamination Certificate
2000-09-11
2004-04-13
Kim, Matthew (Department: 2186)
Electrical computers and digital processing systems: memory
Storage accessing and control
Control technique
C711S112000, C711S114000, C711S100000, C711S154000
Reexamination Certificate
active
06721863
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
This application is based upon and claims the benefit of priority from the prior Japanese Patent Application No. 2000-053814, filed Feb. 29, 2000, the entire contents of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
The present invention relates to a disk control mechanism capable of increasing a speed of a random write with respect to a disk apparatus (a disk memory apparatus) represented by a magnetic disk apparatus in a computer system.
In recent years, there is proposed a log structured file system (LFS) as described in, for example, Jpn. Pat. Appln. KOKAI Publication No. 11-53235 as a technique for increasing the speed of the random write with respect to a disk apparatus in a computer system.
The principle of the log structured file system (hereinafter referred to as LFS) is intended to realize an increase in the speed of the disk write by converting the small block random write to the large block sequential write on the side of the disk control mechanism on the basis of a presupposition peculiar to the disk apparatus to the effect that a large block sequential write of the disk is extremely high in speed as compared with a small block random write. Specifically, data to be written comprising a plurality of small blocks is collected irrespective of its original write position, and is recorded on a disk as a sequential log of one large block with the result that the disk-write speed is increased.
In the case of the application of the LFS, it is necessary to hold information showing a correspondence relationship from the position where a plurality of small blocks of data to be written are supposed to be originally written, namely, the original write position (hereinafter referred to as the original position) intended on the side of a computer to correspond to a position on a log (hereinafter referred to as a log position) and information which stands in a reverse relationship to information showing a correspondence relationship from the log position to the original position. In the foregoing explanation, the former correspondence relationship information is referred to as a forward index while the latter correspondence relationship information is referred to as a reverse index. Furthermore, the both indices are referred to as indices.
The indices are generally held in the computer. As a consequence, when the indices become precarious in the case where the computer comes to a sudden halt because of trouble, the data on the disk unreliable.
Therefore, in the conventional computer system to which the LFS is applied, the indices are held on a dedicated non-volatile memory, for example, an NVRAM (non-volatile random access memory) to provide endurance against damage.
As has been described above, the LFS as disclosed in Jpn. Pat. Appln. KOKAI Publication No. 11-53235 is suitable for realizing an increase in the speed on the random write access to the disk apparatus.
However, the LFS has a problem as described below, and it is important to improve this problem in the practice thereof.
A first problem is that the performance is very likely to be extremely deteriorated with respect to a large block sequential read. The cause thereof is that the large block sequential read has been converted into a small block sequential read as a compensation for the conversion of the small block random write into the large block sequential write. In other words, there is a possibility that the data may be arranged at random at the log position even in the region where the data is continuous at the original position.
A second problem is generated by the application of the LFS to a shared disk in the fail-over system. The fail-over system is a system in which a plurality of computers share the disk apparatus so that even when any computer is damaged, another computer can inherit the processing from the damaged computer. Such a system is referred to as a high availability (HA) system. In this fail-over system, when the primary computer comes to a sudden halt because of trouble, the secondary computer inherits the processing from the primary computer. At this time, the data is handed over to the secondary computer through the shared disk. However, in the LFS, since the index or the like is provided on the non-volatile memory (NVRAM), the data cannot be handed over through the shared disk. In other words, in order to hand over the data, the shared non-volatile memory (NVRAM) becomes necessary.
BRIEF SUMMARY OF THE INVENTION
The present invention has been made in view of the circumstances. An object of the present invention is to provide a disk control mechanism which deals with the rearrangement of data in consideration of the original position of the data the deterioration in the performance of the large block sequential read by the application of the LFS (log structured file system for an increase in the speed of the random write.
Another object of the present invention is to provide a disk control mechanism which eliminates the need of a non-volatile memory and facilitates the inheritance of the indices at the fail-over system by effectively making use of the disk region for the preservation of the indices necessary for the increase in the speed of the random write.
According to a first aspect of the present invention, there is provided a disk control mechanism to which an LFS (log structured file system) is applied wherein data designated by a plurality of disk write requests given from the upper position is collected to be continuously stored in a data block unit having a predetermined size in a region (a log region) which is secured separately from a region (an original region) which can be designated from an upper region on the disk apparatus (namely, which can be seen from the upper position) characterized by comprising rearrangement means for repeating an operation of rearranging the oldest effective data block on the log region at a position on the original region where the data block are supposed to be originally written.
In such a structure, it becomes possible to prevent a reduction in an access performance of a large block sequential read by the application of the LFS because it becomes possible to continuously read the data block from the original region with respect to the read request to a group of data blocks arranged in a continuous region in the original region even if the data blocks are arranged at random at the log position by the rearrangement of the rearrangement means while attempting to increase the speed of a parallel access of the random write (a small block random write) by the application of the LFS.
According to a second aspect of the present invention, there is provided a disk control mechanism to which the LFS is applied, the mechanism being characterized by comprising:
recovery processing means for recovering a forward index in a forward index storage region of a data block secured on a volatile memory, the forward index indicating a correspondence relationship between a position of the data block on the log region and a position on the original region to which the data block is originally written from the reverse index by reading the reverse index from a control block on the log region, namely, at the time of the start-up thereof while allowing the write processing means for collecting data designated by a plurality of disk write requests given from the host device to be provided with a function of adding the control block including the reverse index showing a position on the original region where each of the data blocks which is continuously stored are supposed to be originally written and storing in the log region; and
read processing means for judging which of the log region or the original region the data block designated by the read request is stored by referring to the forward index storage region on the basis of the read request when the read request is given from the host device thereby reading the data block from either the log region or the original region on the basis of the judgment result.
Elmore Stephen
Finnegan Henderson Farabow Garrett & Dunner L.L.P.
Kabushiki Kaisha Toshiba
Kim Matthew
LandOfFree
Disk control mechanism preferable for random disk write 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 control mechanism preferable for random disk write, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Disk control mechanism preferable for random disk write will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3203512