Browse Prior Art Database

Full Track Buffer

IP.com Disclosure Number: IPCOM000100380D
Original Publication Date: 1990-Apr-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 1 page(s) / 31K

Publishing Venue

IBM

Related People

Benhase, MT: AUTHOR [+5]

Abstract

Previous control units have a small buffer to hold a count field and a key field of a record. There is not enough space to hold a data field of an average size record, not to mention to hold more than one record at the same time. As a result, for a channel command that reads consecutive records on a track, control unit retriable errors that occur on the subsequent records being read must be retried by the system. The system has to construct a new read chain to orient the disk device to the record in error and to resume reading the rest of the records on the track. This error recovery scenario is reiterated until all records are read.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 100% of the total text.

Full Track Buffer

       Previous control units have a small buffer to hold a
count field and a key field of a record.  There is not enough space
to hold a data field of an average size record, not to mention to
hold more than one record at the same time.  As a result, for a
channel command that reads consecutive records on a track, control
unit retriable errors that occur on the subsequent records being read
must be retried by the system. The system has to construct a new read
chain to orient the disk device to the record in error and to resume
reading the rest of the records on the track.  This error recovery
scenario is reiterated until all records are read.

      The system performance is improved on a restore operation by
implementing a read track recovery within the control unit.  The
control unit throughput is improved since the same track is reread
without additional device selection and seek operation.

      A buffer large enough to hold all of the records from the track
is added.  Error handling is implemented in the microcode such that
when a control unit re-tryable error occurs on any record being read,
only that error record needs to be reread or to be corrected for
correctable data checks.  All other records read in before the error
record are kept in the buffer and will be retransferred to the system
when the control unit performs retry for the error record.