Sorry if multiple posts (connection died):
If you are using a SAN or RAID array with "read ahead" enabled I find it runs REALLY REALLY faster if I turn it off. "Read ahead" is optimised for large block transfers on sequential SQL SELECTs and does not mix well with small transfers with a random pattern (i.e, hashed files). Of course a HUGE disk cache will offset this to some extent (depends). Regards JayJay -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Doug Dumitru Sent: 09 August 2007 22:27 To: u2-users@listserver.u2ug.org Subject: Re: [U2][UV] Speeding up a big BY.EXP select [EMAIL PROTECTED] wrote: > Sure - but how long does it take on one of your fast new MFT drives ;-) Actually, it is exactly the same speed. The temp files that UV uses to do sorts are accessed linearly so regular arrays work just great. It is random access and updates that kill rotating disks. With Linux it is pretty easy to tell what is going on. 'iostat' can tell you the average IO length and give you an idea of how busy your drives or arrays are. I was kinda surprised that he was taking several hours and my test was 9 minutes. After all, I was testing on 1 80G WD 7200 RPM IDE drive on an AMD Athlon 64 3400 (single core). Not exactly a "high end server". He must have a much larger data set or his server is really busy. --- Doug Dumitru ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/ ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/
