Lee posted on December 6, 2008 11:21

Outstanding read on speed, latency, throughput


I really enjoyed reading this article (link below); it really uses some good analogies to come to conclusions that most of us already know - get a box with as much L1 and L2 cache as possible, and as many spindles as you can for the highest throughput. The disk is obviously the bottleneck still, so beef this area up for maximum performance. Also, controller cache is extremely important. Calculations should be made to understand the size of the reads/writes that SQL Server is performing for a given application, and make sure that sure that controller cache is not smaller than this amount.  Finally, don't mix asynch and synch reads and writes. Enjoy!


What your computer does while you wait
http://duartes.org/gustavo/blog/post/what-your-computer-does-while-you-wait

 

----------------------


Or, optionally, go with Google's model - 700,000 servers with 2 petabytes of RAM.

 


Posted in:   Tags:
blog comments powered by Disqus

by Lee Everest, M.S.

Poll

Do you use Azure or cloud in your organization?



Show Results

Ads

Search


Month List

Calendar

«  February 2012  »
MoTuWeThFrSaSu
303112345
6789101112
13141516171819
20212223242526
2728291234
567891011
View posts in large calendar

Tags

Disclaimer
The opinions, code, examples, et.al. expressed herein are my own personal opinions and do not represent my employer's view in any way, shape form, or fashion.  All code for demonstration purposes - no guarantees, either written or implied, are made.

© Copyright 2012 Lee Everest's SQL Server, etc. weblog