Comments on: Number of servers http://www.unitedwebsite.com/zoltan/2006/11/21/number-of-servers/ Business to Business, Web Development, Import Export News and Comments Tue, 06 Jan 2009 10:56:07 +0000 http://wordpress.org/?v=2.0.4 by: Andrew http://www.unitedwebsite.com/zoltan/2006/11/21/number-of-servers/#comment-56 Thu, 23 Nov 2006 09:34:16 +0000 http://www.unitedwebsite.com/zoltan/2006/11/21/number-of-servers/#comment-56 I'm going to join your TradeHolding.com I’m going to join your TradeHolding.com

]]> by: G http://www.unitedwebsite.com/zoltan/2006/11/21/number-of-servers/#comment-48 Wed, 22 Nov 2006 02:16:26 +0000 http://www.unitedwebsite.com/zoltan/2006/11/21/number-of-servers/#comment-48 "handling large amount of data requires optimal indexes and powerful servers, a lot of RAM especially" Uh, no, not generally. *generally* databases are constrained by disk I/O rather than RAM. *generally* databases benefit most from fast disks (10-15k RPM) in a fast configuration (RAID 0+1 or 6) with a fast controller (128M+ cache). But generalizations may not apply to your environment, and measuring your actual load is the only way to know for sure :) If you find yourself constantly optimizing indexes its time to think about denormalizing your data model, federating your database, or off-loading analytical data from operational data. “handling large amount of data requires optimal indexes and powerful servers, a lot of RAM especially”

Uh, no, not generally. *generally* databases are constrained by disk I/O rather than RAM. *generally* databases benefit most from fast disks (10-15k RPM) in a fast configuration (RAID 0+1 or 6) with a fast controller (128M+ cache). But generalizations may not apply to your environment, and measuring your actual load is the only way to know for sure :)

If you find yourself constantly optimizing indexes its time to think about denormalizing your data model, federating your database, or off-loading analytical data from operational data.

]]>