tag:blogger.com,1999:blog-4516533711330247058.post3936847526475777928..comments2024-03-28T07:32:09.246-07:00Comments on Robert's Db2 blog: DB2 for z/OS Group Buffer Pools: ALLOWAUTOALT, Directory Entries, and GBP SizeRoberthttp://www.blogger.com/profile/02058625981006623480noreply@blogger.comBlogger7125tag:blogger.com,1999:blog-4516533711330247058.post-59456990486232968002020-03-03T21:02:05.996-08:002020-03-03T21:02:05.996-08:00Hello, Henrik.
My short answer to your question i...Hello, Henrik.<br /><br />My short answer to your question is, "I don't know." I have certainly seen, at a number of sites, what you have observed at your site: for some reason, with ALLOWAUTOALT(YES) in effect, the system seems reluctant to increase the size of a GBP, even when that is a possibility (as it is when the structure's max size in the CFRM policy is larger that its allocated size). The system instead seems to strongly favor, as you've noticed, increasing the directory-to-date entry ratio in order to avoid directory entry reclaims - and sometimes that ratio is taken to a very high level. One question is: should you even care about this? I'd check the number of sync read requests due to XI directed to the GBP, per second (sum of sync reads due to XI that returned data, and sync reads due to XI that did not return data, per second). If that value is below 10 per second, I don't much care about even a really low XI read hit ratio. So what if a small number of pages per second have to be read from disk instead of being read from the GBP? If the GBP gets more than 10 sync read requests due to XI per second, I DO care about a really low GBP XI read hit ratio. If that's the case with the GBP on which you're focused, it would seem that you will need to manually increase the size of the GBP in order to boost the XI read hit ratio.<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-87858476398164768982020-03-03T04:37:30.247-08:002020-03-03T04:37:30.247-08:00Hi Robert
One of my group bufferpools has a very h...Hi Robert<br />One of my group bufferpools has a very high directory-to-data entry ratio of 231. The hit ratio for sync read due to XI is very low (below 1%). I found that the allocated pool size is only 225 MB while the max size in the policy specifies 488 MB.<br />We are running with ALLOWAUTOALT = YES. Why does DB2 not increase the group bufferpool size up to 488 MB instead of increasing the directory-to-data entry ratio to an extreme level? Henrik Sand Nielsenhttps://www.blogger.com/profile/02270431365834565035noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-90363654369135684172017-08-09T14:59:45.630-07:002017-08-09T14:59:45.630-07:00Sorry about the delay in responding. The fields of...Sorry about the delay in responding. The fields of interest are QBGLXD (XI GBP synch read, data returned) and QBGLXR (XI GBP synch read, data not returned). They are part of Db2 IFCID 0002, which is generated when statistics trace class 1 is active. The SMF record type is 100.<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-73203520362139894982017-07-31T11:29:50.901-07:002017-07-31T11:29:50.901-07:00Robert, are these values used to calculate the XI ...Robert, are these values used to calculate the XI GBP hit ratio (SYN.READ(XI)-DATA RETURNED and SYN.READ(XI)-NO DATA RETURN) externalized in RMF 74.4 and/or DB2 100 records? Or is Omegamon getting them from internal control blocks? Thanks, Todd HavekostAnonymoushttps://www.blogger.com/profile/00823665541342445097noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-33207171918935625182016-09-08T00:16:14.480-07:002016-09-08T00:16:14.480-07:00Thanks a lot, We have a look.. Appreciate your hel...Thanks a lot, We have a look.. Appreciate your helpNaveenhttps://www.blogger.com/profile/07794927561463907783noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-30106207740443974322016-09-07T12:40:41.227-07:002016-09-07T12:40:41.227-07:00I provided a generalized approach to group buffer ...I provided a generalized approach to group buffer pool sizing, usable for any directory-to-data-entry ratio and any page size, in this blog post:http://robertsdb2blog.blogspot.com/2013/07/db2-for-zos-data-sharing-evolution-of.html<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-6671625911874791812016-09-07T07:32:44.026-07:002016-09-07T07:32:44.026-07:00Hi,
How to determine the size of GBP can be d...Hi,<br /><br /> How to determine the size of GBP can be determined for ration 2:1. In your examples you used 5:1 ratio with that you are diving the total slots by 3. What should be used if the ration is 2:1.<br /><br />ThanksNaveenhttps://www.blogger.com/profile/07794927561463907783noreply@blogger.com