RAID - Placing the data and log files

Question

RAID - Placing the data and log files

Viewing 7 posts - 16 through 21 (of 21 total)

You must be logged in to reply to this topic. Login to reply

andoi SSC Eights! Points: 864 More actions · Answer 1

Another thought

I remember I was at the Visual Studio .Net launch in london and Micro$oft were going on about the performance increases to be had from using a RAW partition for your sql server databases...

Has anyone tried this??

Vladan SSC-Insane Points: 22072 More actions · Answer 2

andoi,

your description almost matches what we were experiencing - I know that situation very well, our disks were at 100% most of the time, too. I agree with your proposal absolutely, in case this isn't enough, you can always think about the second step - going from RAID 5 to RAID 1 0. We did both at once, and the change was so drastical, that soon we began to consider processor and memory upgrade, and finally upped the RAM from 2 to 4 GB with another increase in performance.

Ask the one who wants more memory, what he would say if someone would bring him another loaf of bread while he still didn't finish the first one and has nothing to drink

cbarus SSCommitted Points: 1909 More actions · Answer 3

I am convinced, I still need to have the data/index/log/tempdb on separate arrays even if it is a SAN. I will do some more study on the SAN and if I find anything more than what was discussed here, I will come post here.

Thanks everyone for your time

mybid2003 SSCertifiable Points: 6620 More actions · Answer 4

andoi : Thats the same option I choose (as I mentioned earlier) to increase performance significantly.

Vladan: Thanks for your input. I'll keep your suggestion (RAID 1 0 & Memory) in reserve. Would use when our databases grows larger and gets management's buy in for funding.

Edited by - mdamera on 09/12/2003 12:12:37 AM

.

andoi SSC Eights! Points: 864 More actions · Answer 5

Vladan,mdamera, e.t.c

Out of interest

How many databases, What sizes , and how many transactions per second...

This information would be of some help in proving my case if push comes to shove.. as I appear to work in a CMA environment (As do a lot of us it appears)

I have approx 29 Databases although only 3 or 3 have heavy use.. these are fairly small databases (largest about 1GB).

I currently have @09:07 am

55 user connections

25 transactions /sec(AVE)

45% procesor (TOTAL (2 off)) AVE

3% page file usage AVE

1% Disc Time AVE

These are the best I will See All day...

I will post another set later... A typical one for the afternoon... if anyone is interested..

Vladan SSC-Insane Points: 22072 More actions · Answer 6

We have currently 7 user databases on the server, one of them has around 90GB, the other are small (200-600 MB), have limited number of users (up to 10) and not much traffic. Only the big one has heavy traffic, with almost 100 users logged in during peaks and lots of agents running during the night. BTW, since many of the problems were caused by fragmented indexes, we have a job that runs every night, looks for worst fragmented indexes and runs DBCC INDEXDEFRAG on these.

Great part of data is being imported into the system (e.g. detailed sales of our shops), and then processed - meaning that we experience substantial variability during the day (and night) regarding the usage of CPU, disks etc. Our system has 4GB RAM, 2 processors and 2 administrators

andoi SSC Eights! Points: 864 More actions · Answer 7

Well I think I have got to the bottom of our disk thrashing problem...

The problem did not occur at all on friday..No changes have been made to the server..

The only thing that has changed is that the security settings for our RED DOT content management server have changed and the Database is on the Offending SQL Server..

I am now closely monitoring the server in case this happens again...

Is anyone else using RED DOT CMS and experienced these problems before??