Browse Prior Art Database

Technique for Improving the Utility of Disk Caching

IP.com Disclosure Number: IPCOM000114825D
Original Publication Date: 1995-Feb-01
Included in the Prior Art Database: 2005-Mar-29
Document File: 2 page(s) / 49K

Publishing Venue

IBM

Related People

Annapareddy, NL: AUTHOR

Abstract

Disclosed is a method for improving the utility of a disk cache. Most systems employ a buffer pool in the main memory to store the most recently used pages. The disk subsystems also employ a disk cache for storing the recently requested blocks of data. The disk cache is only effective if there is a cache miss for a block of data in the buffer pool and a cache hit in the disk cache. Since the buffer pool stores almost the same blocks as the disk cache, the effectiveness of the disk cache is not very high since the blocks that miss in buffer pool miss in the disk cache as well. In this disclosure, a method to increase the effectiveness of the disk cache in such situations is presented.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 62% of the total text.

Technique for Improving the Utility of Disk Caching

      Disclosed is a method for improving the utility of a disk
cache.  Most systems employ a buffer pool in the main memory to store
the most recently used pages.  The disk subsystems also employ a disk
cache for storing the recently requested blocks of data.  The disk
cache is only effective if there is a cache miss for a block of data
in the buffer pool and a cache hit in the disk cache.  Since the
buffer pool stores almost the same blocks as the disk cache, the
effectiveness of the disk cache is not very high since the blocks
that miss in buffer pool miss in the disk cache as well.  In this
disclosure, a method to increase the effectiveness of the disk cache
in such situations is presented.

      In present systems, the buffer pool writes out dirty blocks of
data (blocks that are modified and need to be written to disk) to
disk cache.  For deallocating a page from the buffer pool, clean
pages (LRU or Least Recently Used pages) are used.  These clean pages
are discarded and replaced with the new pages being brought into the
buffer pool.  This disclosure proposes that these clean pages being
deallocated from the buffer pool be written to the disk cache and
marked as clean pages in the disk cache.  This effectively links the
two storage spaces (the buffer pool and the disk cache) into a single
storage space.  It also proposes that the only blocks to be stored in
disk cache on a disk access are the block...