xfs
[Top] [All Lists]

xfs on lvm with dm-thin and multipath/multi-volume

To: stan@xxxxxxxxxxxxxxxxx
Subject: xfs on lvm with dm-thin and multipath/multi-volume
From: Dave Hall <kdhall@xxxxxxxxxxxxxx>
Date: Mon, 21 Nov 2016 17:15:41 -0500
Cc: "xfs@xxxxxxxxxxx" <xfs@xxxxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=binghamton-edu.20150623.gappssmtp.com; s=20150623; h=subject:to:references:cc:from:message-id:date:user-agent :mime-version:in-reply-to:content-transfer-encoding; bh=6cbMfxH3zi62AfhybT21+/UHR6B1+A/MRl3RRgnqTjw=; b=z9kWw65mpFbEF1V72ul47i3fyTNId3BUMW/WnA+ApDepOODnc8ubqaT62giyz3Ult1 SDlGFP8qN5o4doSCpJxK1qiOEgZrbRlbSD9WUJ6hSQ3XgOcVBsbgWkewelS0fZU911Fh jBYIsHFuSjNw0URY5esXlaO48iSHSwqvRhabipHf5IVjW/vqm5HvOChGHhLpCPwqztt7 nDrscwzWsXOi1YpIM73o6rgotCZyVNe3RxAJqLxEANadyT8+seBNqSPPVJadNrbI/Dja AvrWRuSVFsGmAby+iKVWebcd2V8P1kQct0/s5Mp0jujK9nsuR3zPabNRqpEyQXvZQaSy bEMA==
In-reply-to: <516CAD27.4070408@xxxxxxxxxxxxxxxxx>
References: <5141C1FC.4060209@xxxxxxxxxxxxxxxxx> <5141C8C1.2080903@xxxxxxxxxxxxxxxxx> <5141E5CF.10101@xxxxxxxxxxxxxx> <5142AE40.6040408@xxxxxxxxxxxxxxxxx> <20130315114538.GF6369@dastard> <5143F94C.1020708@xxxxxxxxxxxxxxxxx> <20130316072126.GG6369@dastard> <515082C3.2000006@xxxxxxxxxxxxxx> <515361B5.8050603@xxxxxxxxxxxxxxxxx> <5155F2B2.1010308@xxxxxxxxxxxxxx> <20130331012231.GJ6369@dastard> <515C3BF3.60601@xxxxxxxxxxxxxx> <51684382.50008@xxxxxxxxxxxxxx> <5168AC0B.5010100@xxxxxxxxxxxxxxxxx> <516C649A.8010003@xxxxxxxxxxxxxx> <516CAD27.4070408@xxxxxxxxxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Icedove/45.4.0
Stan,

It's been a while. I have some new hardware and a chance to do a good XFS layout rather than wonder why previous bad layouts don't perform.

My hardware is a well appointed server with dual HBAs a large SAS-attached array with dual controllers. I'd like to partition 22 spindles (1.8TB each) into multiple small volumes (RAID 5 - 4+1 or RAID 6 - 8+2) with some hot spares and access via multipath. The PVs would be striped or concatenated together with LVM.

I'm wondering about performance with XFS on a dm-thin logical volume and thin snapshots. I'm also concerned about XFS layout in general - RAID stripe alignment and the appropriate number of allocation groups, and if this would be done differently with thin volumes.

The workload is NFS with a very random access pattern and higher than average write activity.

Would you be willing to comment on the current 'state of the art' for all of this? Are there any tools or guidelines for optimal XFS layout?

Thanks.

-Dave

Dave Hall
Binghamton University
kdhall@xxxxxxxxxxxxxx
607-760-2328 (Cell)
607-777-4641 (Office)


<Prev in Thread] Current Thread [Next in Thread>
  • xfs on lvm with dm-thin and multipath/multi-volume, Dave Hall <=