Sign in to follow this  
homer

Matching filesystem block size with RAID stripe size, on SSD RAIDs

Recommended Posts

[Apologies if this seems misplaced. I don't see a forum on RAID configuration.]

I'm configuring a RAID on SSDs. It happens to be 3 drives in a RAID 5, but this is a fairly generic question.

I had the idea that I could reduce stripe read-modify-write operations and write amplification by using a segment size of 4k (which equates to a stripe size of 8k, in my case). Then, I build the filesystem with a block size that matches the stripe size.

The only downside I can see is the overhead of using such a small stripe size, if the controller is too dumb to combine a sequence of 4k reads into fewer, larger reads. The reason I care about performance of small writes is that this filesystem will be used for software builds, among other things. This involves frequently creating large numbers of small/medium-sized files.

From what I can tell, this isn't a very common practice, but I suspect the tendency towards large stripe sizes is a legacy of mechanical disk drives and simple controlers. My RAID "controller" is Linux software RAID (mdadm).

Any thoughts?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this