I was recently asked what changes , if any, we made to the DB2 utilities when we used them to run against our 10 TB fact table during our testing?
First off, to make sure we didn’t run into any issues with DFSORT storage allocations for very large sorts, we did make one slight change to DFSORT’s dynamic storage adjustment (or DSA) feature. The larger your file size, and in our case we were going to sort 9 billion rows or 10 TB of data, the more storage DFSORT will consume.
Continue reading »
Based on a question we received, it was decided that an additional polling question should be created. This question has to do with copying statistics from production to your test system rather than running RUNSTATS on you test system. Unfortunately, what could be some of the most interesting information is difficult to solicit in a poll……. Continue reading »
RUNSTATS is important; I think we can all agree on that. However, how it SHOULD be used and how it’s actually used is still a huge point of discussion. What is extremely cool is that we will get answers to the first part, the “how it SHOULD be used”, from Pat Bossman, an STSM with IBM Silicon Valley Lab’s DB2 for z/OS Query Optimization team. Pat has offered to respond to any of your RUNSTATS inquiries that you post here. This is quite the opportunity for you to get answers to all of those RUNSTATS questions that have been on your mind from an EXPERT on the subject. For the second half of the question, “how it’s actually used”, we’ll attempt to answer by continuing the use of the polling questions to gather RUNSTATS usage information…….. Continue reading »
I guess I’m on a RUNSTATS kick this week. I just made a post on the subject a few days ago, back on September 12, 2011. Today’s however, is more out of curiosity. I’m one of those guys, mostly because of my employer, that assumes everybody pretty much does stuff the way we suggest. Since RUNSTATS can have such a profound and immediate affect on the performance of your SQL, it would be nice to know you thoughts on it. If we get some interesting results or comments, I will bring those forward to development to see if I can get some comments from them that I can post back here……. Continue reading »
Today’s topic started out as a discussion of a DB2 10 feature that I thought was pretty interesting… about a month or so ago. It’s taken this long to get around to completing it. Sometimes timeliness is not my strongest attribute. However, by waiting it has grown to include stuff about zIIP, keyword deprecation, DFSORT DSA, HISTOGRAM, and TABLESPACE SYSTEM…… Continue reading »
An APAR closed recently that corrects an issue that exist with how DB2 collects RUNSTATS HISTOGRAM statistics. This APAR is significant to achieving proper optimization within DB2 10 and DB2 9. It is an APAR that all, regardless of their use of the HISTOGRAM keyword today, should give serious consideration to tracking down and apply. I believe you will be glad you did.
To explain properly why this APAR is significant, Patrick Bossman, part of the DB2 for z/OS Query Optimization team at IBM Silicon Valley Lab,, offered to do a write-up on what this APAR is correcting and why. What follows is his guest bog post….. Continue reading »
It’s kind of a holiday in the US today. Many of us get the Friday after Thanksgiving as a day off. I’m still full from eating too much yesterday so I thought I would relax for a few minutes and knock out a short blog post about a potential issue when collecting index statistics…… Continue reading »