0

We are looking at new servers. Today our SQL Servers are built as follows:

  • C/D on RAID 1 ,
  • E : Tempdb on RAID 1,
  • F:log files on RAID 1 ,
  • G : Catalog data on RAID 1 ,
  • H: model data on RAID 1 ,
  • I : symbols on RAID 1,
  • J: DB maintenance on RAID 1 ,

So we have a lot of SSD’s, all on RAID 1 with no sharing.

We are considering changing to a large RAID 1+0. However the concern is if we will end up with performance impacts since there are random reads that happen throughout the day but not as much as writes.

My question is - whether a large 1+0 RAID is suggested / recommended, which would reduce the quantity of SSD’s that we use and allow us to expand a single container easier or stick with single RAID 1’s?

  • 2
    I'd be much more concerned with what happens to your server if a drive fails. – Erik Darling Jan 30 '19 at 14:08
  • It is typically better (and certainly less management overhead) to spread all data over all disks. So having a large Raid 10 pool is not only easier to maintain, but it will also distribute the load better over all disks and therefore improve performance (at least if your bus is not the bottleneck). – eckes Jun 15 '19 at 15:58
1

RAID level 10(1+0) has been recommended by microsoft for high reads. Refer the doc: https://docs.microsoft.com/en-us/previous-versions/sql/sql-server-2008-r2/ms190764%28v%3dsql.105%29

This level is also known as mirroring with striping. RAID 10 uses a striped array of disks that are then mirrored to another identical set of striped disks. For example, a striped array can be created by using five disks. The striped array of disks is then mirrored using another set of five striped disks. RAID 10 provides the performance benefits of disk striping with the disk redundancy of mirroring. RAID 10 provides the highest read-and-write performance of any one of the other RAID levels, but at the expense of using two times as many disks.

RAID levels higher than 10 (1 + 0) may offer additional fault tolerance or performance enhancements. These levels generally are proprietary systems. For more information about these types of RAID systems, contact the hardware vendor.

| improve this answer | |
-1

Microsoft recommeneds raid 10 for SQL server. but before taking about raid 10 I would like to mention going to RAID 0 for SQL server too risky because if a drive failed in RAID 0 your data will be lost the only way to recover it to use a privous backup file. I would like to recommend you to change RAID configuration as soon as possible.

RAID 1 doesn't support host swap without if you have a fail HDD then you have to shutdown your machine, change HDD, then rebuild your RAID. which is not affordable for production server.

Now let us go back to the original topic using RAID 10 for SQL server, here is the main advantages for RAID 10:

  • there is a disk merroring; if a drive failed then you can swap it and rebuild your RAID without loosing a single byte.
  • you can perform a hot swap; no need to shutdown your server just unplug, plug and rebuild.
  • RAID 10 has an excellent performance and high reliability of both read and write data patterns.

    Here is the main Disadvantages for RAID 10:

  • RAID 10 is a mirrored set of RAID 0 drives. RAID 10 requires a minimum of 4 drives, and can survive the failure of 1 drive. Writes have the performance improvements from RAID 0, along with redundancy from being in a mirror set. Once again, note that the redundancy will mean that you can use only 50% of total disk space. RAID 10 design: enter image description here

Here is some notes from Microsoft about RAID 10 configure RAID 10 Alligned partition on RAID 10

| improve this answer | |
  • "if a drive failed in RAID 1 your data will be lost" - what do you mean? Raid 1 mirrors data, so it does tolerate single disk loss. – vonPryz May 8 '19 at 7:25
  • my bad that is raid 0. its a typo. – Ahmad Abuhasna May 13 '19 at 5:57
  • Most advanced raid controllers also allow hotswap for raid1 – eckes Jun 15 '19 at 15:59
  • @eckes true, unless drive is cabled. and nearly all medium to large servers doesn't use RAID 1. – Ahmad Abuhasna Jun 16 '19 at 7:38
  • There are ways around the classical boot/system drive, but if you prefer booting your servers off of them a raid1 is a classical usecase. I guess only making servers „stateless“ brought that out of fashion, you just replace the whole unit when it’s System drive failed. But it is the cheapest raid option if a single disk size is enough. – eckes Jun 16 '19 at 17:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.