Posts of Interest
I don't make a habit of just posting links as I think it's like a form of cheating, however sometimes there are things I want to keep track of and I know If I put the link on my blog I'll find it again!
so: a short one here; We've just bought some new servers with 10 core processors, or 20 if you turn on hyperthreading. That's 80 cores/threads which is sort of a waste if you're not going to make use of parallel plans, however it does make a backup fast! I saw this post form the CSS team and thought - hmmm I must remember this http://blogs.msdn.com/b/psssql/archive/2011/09/01/sql-server-2008-2008-r2-on-newer-machines-with-more-than-8-cpus-presented-per-numa-node-may-need-trace-flag-8048.aspx btw I think he means cores not cpus.
Now for an interesting one indeed: I've always been one to want to designate indexes/keys descending when tables have incremental counters as a key - or the key on a secondary index is a date and I know we usually pick the latest dates, as an example. So my concern has always been that if I want the latest rows from a table I should have descending indexes; I seem to recollect somewhere in my distant past that this was actually important on a RDBMS ( not sql server ). So it's something i've been doing on and off for many years - the problem being that if asked to justify why i want to make an index descending could I prove that it made a difference?
Well earlier this year I attempted to prove that it did in fact make a difference ( sql 2008 and sql 2008 R2 ) but my attempts failed and other than it just making a select top xxx always return the latest rows I wasn't able to show a difference. Well today I found out how poor my attempts to show this were < grin > and I bow down to Fabiano Amorim who has an article in simple-talk which illustrates what I was trying to achieve. http://www.simple-talk.com/sql/database-administration/statistics-on-ascending-columns/?utm_source=simpletalk&utm_medium=email-main&utm_content=ascending-20110905&utm_campaign=SQL