Re: Distributions

In article <361E6464.7FF4A847@weideneder.de>, Stefan Weideneder <stefan@weideneder.de> writes >What are the costs for an UPDATE STATISTICS MEDIUM/HIGH ? We need >at least 12 bytes for each bin ( for an integer field ), let's say >we have an average of 20 bytes. If we would have about 100 bins per >column ( resolution 1 ), this would cost 1,200 bytes per column. >Let's assume a table is made up of appr. 10 columns, this would >cost about 12,000 bytes per table. Now if we have 5,000 tables >( in a BaaN environment we have about 8,000 to 40,000 tables ) >this would cost about 60MB of memory. Maybe your systems have ?? This is 60Mb of disk space. They are NOT all run in parallel at the same time. Our small Sun box can manage 2Mb/s disk I/O so we are talking 30 seconds of disk I/O! Beside you can run them overnight or at different times e.g. 1000 per day. >enough memory and fast CPUs which can manage the evaluation >of the distributions, but a lot of customers do not have such >machines. > You talk about 5000 tables and then small machines in the same paragraph?? Who runs 5000 tables on small machines?? >Finally, I would use UPDATE STATISTICS HIGH, if the optimizer >would generate a wrong query plan with UPDATE STATISTICS MEDIUM. >And I would avoid an UPDATE STATISTICS MEDIUM as long as the >optimizer generates the correct query plan with an ordinary >UPDATE STATISTICS LOW.> >Bye > >Stefan Weideneder -- David Williams Maintainer of the Informix FAQ Primary site (Beta Version) http://www.smooth1.demon.co.uk Official site http://www.iiug.org/techinfo/faq/faq_top.html I see you standin', Standin' on your own, It's such a lonely place for you, For you to be If you need a shoulder, Or if you need a friend, I'll be here standing, Until the bitter end... So don't chastise me Or think I, I mean you harm... All I ever wanted Was for you To know that I care