new SQL server 2012 and Raid Sets

Hi Experts,

I want to buy new SQL servers.
I want to isolate the data with different raid sets.

Which Raid is recommended when I go for SQL2012 or 2014 ?

I want to separate :

OS
DATA Files
LOG Files
TEMPDB
Backup

Which has the best perfomance ?
Eprs_AdminSystem ArchitectAsked:
Who is Participating?

[Webinar] Streamline your web hosting managementRegister Today

x
 
ste5anConnect With a Mentor Senior DeveloperCommented:
That's basically ok, when it fits into your budget and the anticipated workload.

OS: Remember that per default SQL Server installs master, msdb and also model on OS.
DATA/LOG: depends on the recovery mode. LOG needs to be fast. Save only when you run in full recovery mode.
TEMPDB: SSD's are fine.
BACKUP: Depending on the size of the data and the recovery mode (aka number of necessary full backups) having it fast is also important.
0
 
ste5anConnect With a Mentor Senior DeveloperCommented:
To separate those on different volumes on different spindles make sense.

Which has the best perfomance ?

This depends on your requirements and the number of disks you can afford. Here's an short overview: RAID Levels and SQL Server.

The important point is: Do you want redundancy (RAID 5/6EE) or performance (RAID 1+0)?

Depeding on your load and the processor architecture (NUMA or not, numbers of physical cores), a hugh performance boost can be achieved by organizing tempdb correctly (SSD's instead of disks, multiple file groups per phys. core).
0
 
Eprs_AdminSystem ArchitectAuthor Commented:
Hi,
I want perfomance and with RAID10 I also have redundancy, when I configure my monitoring.
0
Creating Active Directory Users from a Text File

If your organization has a need to mass-create AD user accounts, watch this video to see how its done without the need for scripting or other unnecessary complexities.

 
ste5anConnect With a Mentor Senior DeveloperCommented:
Just keep in mind, that RAID 10 means to mirror one strip-set to  a second one or more. The number of disks which may be defect at a time, but still have a working raid is [number of mirrors] - 1. So using 8 disks you have to possible balanced RAID10 configurations:

1. 2 mirrors of each 4 disks
2. 4 mirrors of each 2 disks

In the first case only one drive may fail. An further failing drive may be on the second mirror and lead to a failed RAID. Thus you have 1:8.
In the second case three drives may fail. This still leaves one mirror intact. Thus we have a ratio of 3:8.  

So redundancy depends on the configuration.

But those two configrations will have a different performance. But this is a kind of stocastical consideration depeding on some hardware factors of the system (block size, number of controllers, write cache, etc.) which also needs some testing with the actual hardware. The background is quite simple, a write or read is completed, when one mirror has completed this operation. Thus normally having more disks per mirror means better performance. But as already said, this depends strongly on some HW facts.
0
 
Eprs_AdminSystem ArchitectAuthor Commented:
I plan with this Raid configuration for my SQL Server:

Each Raid set is physically separated.

OS - Raid 1 with SATA 15k disks
DATA - RAID 10 with SATA 15k disks
LOGS - RAID 10 with SATA 15k disks
TEMPDB - RAID 1 with SSD´s
BACKUP -RAID 1 with SATA 15k disks
0
 
Eprs_AdminSystem ArchitectAuthor Commented:
ok then I can ask for some quotes from different vendors.
0
All Courses

From novice to tech pro — start learning today.