tag:blogger.com,1999:blog-4516533711330247058.post3694894477530865841..comments2024-03-29T05:35:42.451-07:00Comments on Robert's Db2 blog: DB2 for z/OS ZPARMs that Organizations Consistently Set in a Sub-Optimal Fashion Roberthttp://www.blogger.com/profile/02058625981006623480noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-4516533711330247058.post-60987292867634411672019-02-11T11:11:55.496-08:002019-02-11T11:11:55.496-08:00OK Robert. When we update the db2 we will try Huff...OK Robert. When we update the db2 we will try Huffman compression.<br />thank youElvio Comunellohttps://www.blogger.com/profile/05032068097599126397noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-1252752877138866072019-02-08T19:45:35.306-08:002019-02-08T19:45:35.306-08:00OK, with HONOR_KEEPDICTIONARY set to NO, when a RE...OK, with HONOR_KEEPDICTIONARY set to NO, when a REORG changes a table's row organization from BRF to RRF then a new compression dictionary will be built, regardless of whether or not KEEPDICTIOONARY is specified on the REORG job's control statement.<br /><br />It seems that the change in physical column order resulted in a situation in which the data simply could not be compressed as it had been before. APAR PH04424 indicates that Huffman compression (versus the Ziv-Lempel compression that Db2 for z/OS has long used) will be available at some point in a Db2 12 environment - perhaps when that compression capability is available it will provide you with a better result for this particular table space.<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-79027508281139104702019-02-08T12:21:13.489-08:002019-02-08T12:21:13.489-08:00Thanks Robert for the quick response.
The parame...Thanks Robert for the quick response. <br />The parameter HONOR_KEEPDICTIONARY is set to NO but in the reorg we do not use the keepdictionary relying on the db2 to optimize the compress.<br />Elvio Comunellohttps://www.blogger.com/profile/05032068097599126397noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-15659053313469869982019-02-08T11:01:24.850-08:002019-02-08T11:01:24.850-08:00In my experience, this result - a reduced level of...In my experience, this result - a reduced level of compression as a result of going from BRF to RRF - is unusual but not unprecedented. The reduced compression ratio is likely due to the change in data-value patterns in rows resulting from relocation of varying-length columns to the end of each row. Do you have the ZPARM parameter HONOR_KEEPDICTIONARY set to NO or YES?<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-56855191280700814072019-02-08T06:54:35.280-08:002019-02-08T06:54:35.280-08:00Hi Robert, thanks for the clear explanation.
We mi...Hi Robert, thanks for the clear explanation.<br />We migrated from v11 to v12 and the parameter of the depreciation zpam, so when reorganizing we changed from BRF to RRF.<br />But there was an undesired consequence, in several compress tables the allocated space increased a lot. <br />We note that they are UTS PBR and the table has the most fields VARGRAPH and few DECIMAL.<br />For example, on the one hand with BRF the Average row lengths is 471 and the Pct of pages saved by compression 76 and on the other hand with RRF the Average row lengths 863 and the Pct of pages saved by compression 57<br /><br />Can this difference be explained and how can we achieve the previous compression with RRF?<br /><br />ThanksElvio Comunellohttps://www.blogger.com/profile/05032068097599126397noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-61300361646384760642018-07-18T08:15:07.057-07:002018-07-18T08:15:07.057-07:00RRF is short for reordered row format, and I am no...RRF is short for reordered row format, and I am not familiar with the use of the term "extended" in relation to RRF (the alternate row format is BRF, which stands for basic row format). In any case, when the Db2 catalog was changed in going to Db2 10 enable new-function mode (via the CATENFM utility), the catalog tables were changed to RRF (I say "were" because I assume you're on Db2 11 or Db2 12).<br /><br />RobertRoberthttps://www.blogger.com/profile/02058625981006623480noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-64481661065798628482018-07-16T05:55:33.960-07:002018-07-16T05:55:33.960-07:00Thanks for sharing. Is the db2 catalog also in RRF...Thanks for sharing. Is the db2 catalog also in RRF=EXTENDED Format, or should this be done during a catalog reorg ?MM Meine Meinunghttps://www.blogger.com/profile/12994386219289115967noreply@blogger.comtag:blogger.com,1999:blog-4516533711330247058.post-51334841953467444842016-12-01T12:52:01.528-08:002016-12-01T12:52:01.528-08:00Great information. Thank you for sharingGreat information. Thank you for sharingtroycolemanhttps://www.blogger.com/profile/10589139698593456037noreply@blogger.com