• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

SQL - RAID Levels - RAID Size etc

Jay_2

2[H]4U
Joined
Mar 20, 2006
Messages
3,583
Hi!!

I am looking at building a new SQL server

We are looking at 2 x 15k 146GB SAS disk for OS and 14 x 600GB 15k SAS for databases but I am not sure what the best RAID level would be for this setup. I know that SQL loves the spindles but at RAID 10 we will reduce the usable disks in half, RAID 5 would be terrible to rebuild at this size and also parity would slow down the writes.

What RAID level would you use on this and why?
 
Generally raid10 is still king for disk I/O for SQL as it usually entails a lot of small random writes (the write hole that parity raid is usually bad with).

You can still usually get some pretty respectable I/O levels even with raid6 but it really depends on your database I would say. If you have a very I/O heavy DB (which I would guess from the massive number of disks) then raid10 is likely always going to be the best performer for SQL.

On my own colo'd server with areca I just made a small volume set which is raid10 for MySQL and raid6 for pretty much everything else:

Code:
# cli64 vsf info
  # Name             Raid Name       Level   Capacity Ch/Id/Lun  State
===============================================================================
  1 OS RAID6         8x1.5TB RAID SETRaid6     64.0GB 00/00/00   Normal
  2 DATABASE RAID10  8x1.5TB RAID SETRaid1+0   30.0GB 00/00/01   Normal
  3 DATA RAID6       8x1.5TB RAID SETRaid6   8891.0GB 00/00/02   Normal
  4 SSD RAID0        4x OCZ Core v2 SRaid0   1032.0GB 00/00/03   Normal
===============================================================================

It was a nice way to get around the small writes using more I/O due to raid6 and still allow for efficient space usage for the rest of the machine. But if your machine is dedicated to sql then definitely raid10 all the way.
[/code]
 
How much data do you have in the SQL database(s)? How is it organized: one large table, or many independent tables? How much of your dataset is hot? How much expected growth do you have? Depending on the layout, it might be worth switching to 7k disks with SSD for indices or something like that.
 
On all of our SQL servers at work we use RAID 1 for OS RAID 10 for Database and LOGS on seperate RAID sets. Also we have a seperate RAID 10 set for the TempDB
 
I believe Microsoft recommends RAID 10 for the DB and RAID 1 for log files. Their reasoning likely applies to other OSes/SQL variant.
 
@OP

One thing to consider is that MS SQL Server can also perform well using multiple RAID 1 pairs; seemingly there's a way to spread the tables among the spindles.

Read this for more info.
 
Back
Top