Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Sort Process for Migration - Storage to Optical Versus Tape

IP.com Disclosure Number: IPCOM000106270D
Original Publication Date: 1993-Oct-01
Included in the Prior Art Database: 2005-Mar-20
Document File: 4 page(s) / 107K

Publishing Venue

IBM

Related People

Goldfeder, ME: AUTHOR [+3]

Abstract

Tape and optical devices and libraries receive data migrated from high-end DASD. This migration occurs when the cost of the high-end DASD cannot be justified for less frequently accessed data.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Sort Process for Migration - Storage to Optical Versus Tape

      Tape and optical devices and libraries receive data migrated
from high-end DASD.  This migration occurs when the cost of the
high-end DASD cannot be justified for less frequently accessed data.

      The problem is deciding whether to migrate data to tape or to
optical.  Streaming tape has inherently faster effective data rates
than optical, which makes migrating to tape the right decision for
data which has historically been accessed sequentially.  However, the
fast seek time of the optical read/write head is orders of magnitude
faster than seeking on tape, which makes optical the right choice for
data which has tended to be accessed in a random fashion.

      The tracking of whether accesses to data sets on DASD have been
sequential or random is proposed.  This history will then be used to
decide whether to migrate to tape, for sequentially accessed data, or
optical, for non-sequentially, i.e., randomly, accessed data.

      There are known sequential detection algorithms used today to
manipulate DASD I/O.  Also, the DASD application software often
specifies sequential access via a "define extent bit."  Therefore, it
is possible to know whether a given access is sequential in nature.
We would use such an algorithm in Fig. 1, step 12.

      Further proposed is a scoring algorithm be used, with +1
awarded for a sequentially accessed data set (step 13) and -1 for a
randomly accessed data set (step 14).  The score would be awarded
when the data set was closed, step 15, to avoid the random access of
data from "running up" a large negative score.  This score could be
cumulative over the data set's entire history, or it could be a
running average of the las...