HP 3000 Manuals

ALLBASE/SQL Performance and Monitoring Guidelines

Using Parallel Serial Scans 

When you need to sequentially read a large table, you can improve
performance by using a parallel serial scan.  The ALLBASE/SQL optimizer
uses a parallel serial scan when it is able to prefetch pages from
multiple disk drives in parallel.  A parallel serial scan is useful only
for large tables that must be read sequentially.  It is not beneficial
for small tables or for tables that are accessed only by an index.

To take advantage of a parallel serial scan, the following conditions
must be met:

   *   The DBEFileset containing the table must contain multiple
       DBEFiles.

   *   The DBEFiles containing the table must be placed on separate disk
       drives.  You can specify the disk drive with the DEVICE clause of
       the CREATE DBEFILE statement.  If the DBEFile has already been
       created, you can use the SQLUtil MOVEFILE command.

   *   The DBEFiles cannot be pseudo-mapped because the file system's
       prefetching must not be bypassed.

   *   The DBEFiles must not contain pages from other tables or index
       pages from the same table.  Put other tables in different
       DBEFilesets.  Use separate INDEX and TABLE DBEfiles for the index
       and data pages of the table.
[REV END]

MPE/iX 5.0 Documentation