ca.portal.admin

Different flavors of relational database

Discussion created by ca.portal.admin on May 2, 2006
Latest reply on May 2, 2006 by ca.portal.admin
We did a comparison retrieval between IDMS/SQL (on the mainframe) and SQL
Server (on a Windows server). A database was defined the same way on both,
and loaded with the same data. Queries that used keys (index or calc) had
about the same response time on both. But queries that generated an area
sweep ran noticeably faster on SQL Server, for example
1 second vs. 15 seconds. I am trying to figure out why such a big
difference.

I did not consider the response time for area sweep unreasonable for the
size of the database. But the programmer insisted that it was, and proved
his point with the SQL Server demo. I considered these queries to be large
(for online access), but the SQL Server DBA said that they are considered
small by SQL Server standards. They reported that DB2 (on AIX) could also
do similarly large queries with subsecond response time.

I do realize that there are many differences between the environments, i.e.
hardware, software, operating system, DASD, number of users, etc.
But is there something inherently different about IDMS/SQL that would cause
large queries (area sweeps) to run so much slower than the other two
relational databases?

If anyone else has done similar comparison testing, would you be willing to
share your results?

Kay Rozeboom
State of Iowa
Information Technology Enterprise
Department of Administrative Services
Telephone: 515.281.6139 Fax: 515.281.6137
Email: Kay.Rozeboom@Iowa.Gov

"
IDMS Public Discussion Forum
IDMS-L@LISTSERV.IUASSN.COM
SMTP
IDMS-L@LISTSERV.IUASSN.COM
IDMS-L@LISTSERV.IUASSN.COM
SMTP








Normal

Normal
Re: Different flavors of relational database
"I would begin by looking at the SMF data to nail down how much time was
actually spent in IDMS. Different network paths to the data sources can
add incredible amounts of the elapsed time of a transaction. We've had
customer web applications that develop and test on the DB2 LUW
environment and then complain about elapsed times when they move to
z/OS. When I look at the SMF data the in-DB2 times are about the same
on z/OS as the were on LUW but the application is now going through
several firewalls and competing for network resources with the rest of
the production workload.

Outcomes