[Webinar] Streamline your web hosting managementRegister Today

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 644
  • Last Modified:

Metalun vs. larger raid group

I was recently asked a question I cannot answer and I'd like to get some outside input on this.
We have an EMC Clariion environment.

We have a server that requires a single 2tb lun and pretty high IOPS, more than a single 5disk raid5 raidgroup can provide.

I have two choices, either go with a 10disk raid5 raidgroup with a single 2tb lun or go with 2x 5disk raid5 raidgroups and create a 2tb metalun across them.

Both would be utilizing all spindles. Every other raidgroup we have is a 5disk raid5 group. Other than "trying to keep all the raidgroups the same" what other reason do I have to go with one over the other?
0
narcus
Asked:
narcus
  • 3
  • 3
1 Solution
 
narcusAuthor Commented:
I'd like to add that I am referring to a striped metalun, not concatenate
0
 
Duncan MeyersCommented:
The 10 disk RAID 5 group will be marginally faster, however that depends on your I/O profile. 4+1 and 8+1 RAID groups have an edge in some workloads as they allow the CLARiiON to perform "square" back-end writes - this effect will be most noticeable with a workload that generates heavy sequential writes. However, I recommend configuring the RAID groups as two 4+1 RAID 5 groups and use create a 2TB metaLUN out of two 1TB components as this allows for simpler expansion down the track. On the other hand, you could also consider a 10 disk RAID 1/0 group if you have high performance requirements, providing that gives you enough disk space. A 10 disk RAID 1/0 group of 450GB drives will give you 2016GB usable.
0
 
narcusAuthor Commented:
Thanks for the answer! We don't have insane IO requirements for this, but it DOES require 2tb which is more than one 4+1 can provide so I was left with the choice between the two. While sucking every bit of IOPS out of the array isn't necessary, I also don't want to waste performance. It sounds like it won't really make a difference with either setup. All of our other raidgroups are 4+1 and so far, keeping it homogeneous is the only compelling reason I have to go with a 4+1 over 8+1.
0
Hire Technology Freelancers with Gigs

Work with freelancers specializing in everything from database administration to programming, who have proven themselves as experts in their field. Hire the best, collaborate easily, pay securely, and get projects done right.

 
Duncan MeyersCommented:
A 4+1 RAID 5 group also rebuilds faster in case of a catastrophic disk failure - that's probably enough to tip it over to 4+1s.

(Depending on what FLARE code you're running, your array will likely have pro-active hot-sparing where the array kicks in a mirror operation to the hot spare if a drive shows signs of failure. The pro-active hot spare operation typically takes about 30 minutes for a FC drive, compared to about 4 hours for a rebuild from parity, so you are less exposed to double drive failure and performance will not be clobbered)
0
 
narcusAuthor Commented:
That's a pretty significant reason. So far, out of all my google searching, this has been the first GOOD reason to do one over the other.
0
 
Duncan MeyersCommented:
Thanks! Glad I could help.
0

Featured Post

Upgrade your Question Security!

Your question, your audience. Choose who sees your identity—and your question—with question security.

  • 3
  • 3
Tackle projects and never again get stuck behind a technical roadblock.
Join Now