RAID configuration for optimal performance
Hi, Does anyone know what type of RAID is proposed by informix? To be more precise, is it possible to achieve the same performance as RAID 1+0 (mirroring + stripping) with RAID 5? Is it true that informix suggests that we do not use RAID 5? Any references will be greatly appreciated.... Thanks -- Ioannis Demetriades CTL Ltd.
"Ioannis Demetriades" <ioannis@ctl.com.cy> writes: > Hi, > > Does anyone know what type of RAID is proposed by informix? To be more > precise, is it possible to achieve the same performance as RAID 1+0 > (mirroring + stripping) with RAID 5? Is it true that informix suggests that > we do not use RAID 5? RAID 5 sucks, check out Art's raving on this group with www.deja.com... RAID 10 and more disks than you can imagine is the answer:) > Any references will be greatly appreciated.... www.deja.com http://x28.deja.com/[ST_rn=ps]/getdoc.xp?AN=555309930.1&CONTEXT=948710239.1594425351&hitnum=16 Thomas
Ioannis Demetriades wrote: > > Hi, > > Does anyone know what type of RAID is proposed by informix? To be more > precise, is it possible to achieve the same performance as RAID 1+0 > (mirroring + stripping) with RAID 5? Is it true that informix suggests that > we do not use RAID 5? Informix does not care what you use they support anything. > Any references will be greatly appreciated.... As Thomas suggests scan the IIUG archives for my own postings on the subject. I usually 'let' myself be provoked into a tirade on the the evils of RAID5 each summer. Obnoxio, I'll admit that an EMC or Sun A5000 array will perform acceptably in RAID5 format, yes sometimes even for writes simply because Informix's cache and the array controller's HUGE caches seem to hide the sluggishness of the writes, but I cannot agree that a properly configured RAID10 array does not blow it away and with less cache$ both in the engine and on the controllers. Proof? Look at the checkpoint times with both NOT the application throughput which is being improved by cache. BTW Thomas they're more of a rant than a rave 8-) Art S. Kagel
"Art S. Kagel" <kagel@bloomberg.net> writes: [SNIP] > BTW Thomas they're more of a rant than a rave 8-) > > Art S. Kagel Ach, rant-party then? Thomas
How would you prove that the performance of a RAID 5 is not as good as RAID 1+0 to your client? Do you know of any official documents that address the performance of RAID 5 as compared with other levels of RAID Thanks for your replies. Ioannis Demetriades <ioannis@ctl.com.cy> wrote in message news:388c284e.0@news.spidernet.net... > Hi, > > Does anyone know what type of RAID is proposed by informix? To be more > precise, is it possible to achieve the same performance as RAID 1+0 > (mirroring + stripping) with RAID 5? Is it true that informix suggests that > we do not use RAID 5? > > Any references will be greatly appreciated.... > > Thanks > > -- > Ioannis Demetriades > CTL Ltd. > >
"Ioannis Demetriades" <ioannis@ctl.com.cy> writes: > How would you prove that the performance of a RAID 5 is not as good as RAID > 1+0 to your client? Do you know of any official documents that address the > performance of RAID 5 as compared with other levels of RAID By understanding what RAID 5 is? There's probably some benchmarks out there... This one is close:) -RAID 1 & RAID 10 vs. RAID 5 In A High Performance NFS Serving Environment "RAID 5, while being the most efficient in its use of disk resource, will not deliver optimal performance. RAID 10 will deliver 4X the write IOPS, and about the same read IOPS. " http://www.maxstrat.com/1_10v5_wht.html Adaptec has some on this (basic stuff): http://www.adaptec.com/technology/whitepapers/raid.html http://www.adaptec.com/products/guide/abcraid.html A nice series of articles in SunWorld: http://www.sunworld.com/swol-06-1999/swol-06-raid1.html http://www.sunworld.com/sunworldonline/swol-07-1999/swol-07-raid2.html http://www.sunworld.com/sunworldonline/swol-08-1999/swol-08-raid3.html http://www.sunworld.com/sunworldonline/swol-09-1999/swol-09-raid4.html http://www.sunworld.com/sunworldonline/swol-10-1999/swol-10-raid5.html Thomas