tag:blogger.com,1999:blog-4516533711330247058.post5012861785231204149..comments2024-03-28T07:32:09.246-07:00Comments on Robert's Db2 blog: DB2 for z/OS: How I Use -DISPLAY BUFFERPOOL OutputRoberthttp://www.blogger.com/profile/02058625981006623480noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-4516533711330247058.post-74832394086418201442016-04-20T15:59:25.083-07:002016-04-20T15:59:25.083-07:00Sorry about the delay in responding, Ilya.
You co...Sorry about the delay in responding, Ilya.<br /><br />You could try reducing VDWQT to the default value of 5, versus 10. That should trigger more deferred write activity, and the result could be more available buffers for receiving pages brought into memory via prefetch.<br /><br />You indicate that you can't make the buffer pool any larger because of a lack of real storage in the z/OS LPAR. Could you make the buffer pool larger by decreasing the size of another buffer pool (or pools) by a like amount (so that the overall size of the buffer pool configuration would remain the same)? Check the total read I/O rate (all synchronous reads plus all prefetch reads, per second) for each buffer pool allocated for the DB2 subsystem. Do you see that one or more pools has a low total read I/O rate (less than 100 per second)? If so, consider taking buffers from that pool (or pools) by reducing VPSIZE, and adding the same number of buffers to the pool in question (the one for which you're seeing lots in the way of PREFTECH DISABLED - NO BUFFER).<br /><br />Longer term, push for more memory on the system. That's a way to boost throughput and CPU efficiency for DB2-accessing applications, without increasing software license costs (since z Systems software is not licensed on a server memory basis).<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-22217444623490531022016-04-13T06:18:30.255-07:002016-04-13T06:18:30.255-07:00Hi Robert,
Thank you for this article - very in...Hi Robert,<br />Thank you for this article - very informative (as usual). <br /> I have a question regarding Prefetch disable - no buffer condition. <br /> I recently noticed that one of my buffer on specific LPAR ( we have 8-way sharing) getting thousands and thousands Prefetch-disable- no buffer condition. In fact, for some interval , it's reaching over 100K such occurrences per hour. <br />The buffer was at 110K pages . I increased it to 150K pages, but don't see much of the positive effect. There is some reduction in number of Prefetch-no-buffer condition , but not substantial. <br /> This buffer is used for index datasets only, VPSQT is set at 85% ( vdwqt is at 10 , and dwqt is at 30%). <br /> I cannot increase this BP anymore ( no more available storage on LPAR ) . <br />Could you recommend anything else to address this issue ? <br /> Regards Ilya Anonymoushttps://www.blogger.com/profile/16051912882874637061noreply@blogger.com