19.06.2013 Views

DB2 UDB for z/OS Version 8 Performance Topics - IBM Redbooks

DB2 UDB for z/OS Version 8 Performance Topics - IBM Redbooks

DB2 UDB for z/OS Version 8 Performance Topics - IBM Redbooks

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>DB2</strong> V8 in new-function mode increases the number of active log data sets from 31 to 93,<br />

allowing <strong>DB2</strong> to keep more log data online and there<strong>for</strong>e available <strong>for</strong> any recovery.<br />

In addition, <strong>DB2</strong> V8 in new-function mode increases the number of archive log data sets <strong>DB2</strong><br />

remembers from 1,000 to 10,000. This increases the amount of log data <strong>DB2</strong> is able to<br />

access and use in recovery, there<strong>for</strong>e satisfying the larger window often required by Service<br />

Level Agreements (SLA).<br />

4.13.2 Up to 65,041 open data sets<br />

<strong>DB2</strong> V8 increases the number of open data sets from 32k to 65,041. This enhancement is<br />

available starting with z/<strong>OS</strong> V1.5 base code.<br />

Open data sets require storage <strong>for</strong> MVS control blocks below the 16 MB line. This has been<br />

the cause of virtual storage constraint problems below the 16 MB line in the past. Dynamic<br />

allocations can request these control blocks to be allocated above the 16 MB line. With z/<strong>OS</strong><br />

V1.5, media manager, the interface that <strong>DB2</strong> uses, can fully exploit this feature. This reduces<br />

the amount of storage that <strong>DB2</strong> requires below the 16 MB line.<br />

This enhancement does not only provide virtual storage relief. It also removes a potential<br />

scalability problem with growing systems and growing machine capacity.<br />

In addition, <strong>DB2</strong> V8 no longer uses system-generated DDNAMEs, as the current limit there is<br />

also 32k. <strong>DB2</strong> generates its own DDNAMEs <strong>for</strong> dynamic allocation. This way <strong>DB2</strong> can have<br />

better open data set per<strong>for</strong>mance as well, and, can avoid the problem of the task that<br />

generates the system DDNAMEs becoming a bottleneck <strong>for</strong> parallel data set open.<br />

When DSMAX is reached, <strong>DB2</strong> currently attempts to close up to 3% of the open data sets.<br />

However, 3% of 65,041 is still a large number, and this burst of data set close activity could be<br />

disruptive. There<strong>for</strong>e <strong>DB2</strong> now closes MIN(3%,300) data sets when DSMAX is reached.<br />

PFT UQ96862 <strong>for</strong> APAR PQ96189 is related to the number of open data sets.<br />

4.13.3 SMF type 89 record collection enhancements<br />

SMF type 89-1 interval records provide <strong>IBM</strong> with the required “Software Usage Report” <strong>for</strong><br />

reporting on zSeries Software License Requirements.<br />

Prior to V8, <strong>DB2</strong> automatically used detailed tracking of measured usage if SMF type 89<br />

records were activated.<br />

<strong>DB2</strong> V8 provides a new DSNZPARM parameter, SMF89, which lets you specify whether <strong>DB2</strong><br />

is to do detailed tracking <strong>for</strong> measured usage pricing. The default value is NO, which means<br />

that <strong>DB2</strong> does not do detailed measured usage tracking. If the SMF type 89 record is<br />

activated, only high-level tracking is recorded in the SMF type 89 record.<br />

If you select YES, <strong>DB2</strong> does detailed measured usage tracking if SMF type 89 records are<br />

activated. When SMF89 is set to YES, <strong>DB2</strong> invokes a z/<strong>OS</strong> service on every entry or exit into<br />

or out of <strong>DB2</strong> to ensure accurate tracking. A new SMF type 89 record is cut on each SQL<br />

statement.<br />

If you select NO, <strong>DB2</strong> CPU is reduced; however, the amount of time spent in <strong>DB2</strong> as<br />

measured by SMF 89 is increased.<br />

A "typical" CPU overhead <strong>for</strong> <strong>DB2</strong> per<strong>for</strong>ming detailed tracking of CPU usage is in the range<br />

of 0 to 5%. Since there is a fixed overhead per SQL statement, the overhead tends to be<br />

higher <strong>for</strong> short-running SQL statements and negligible <strong>for</strong> long-running SQL statements. For<br />

Chapter 4. <strong>DB2</strong> subsystem per<strong>for</strong>mance 213

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!