| To: | <david@xxxxxxxxxxxxx>, <aelder@xxxxxxx> |
|---|---|
| Subject: | Re: [RFC, PATCH 0/2] xfs: dynamic speculative preallocation for delalloc |
| From: | <Ivan.Novick@xxxxxxx> |
| Date: | Thu, 14 Oct 2010 17:50:15 -0400 |
| Accept-language: | en-US |
| Acceptlanguage: | en-US |
| Cc: | <xfs@xxxxxxxxxxx> |
| In-reply-to: | <20101014211626.GD4681@dastard> |
| Thread-index: | Actr6QTXYhXaG4P6Q2Wijr33fHtHEgAAMPs2 |
| Thread-topic: | [RFC, PATCH 0/2] xfs: dynamic speculative preallocation for delalloc |
>> The reported numbers were a drop in read throughput of ~15% on a >> GB/s class filesystem when the files interleaved. I have seen ~50% performance improvement in read rate when changing from small extents to large extents with XFS. Essentially going from not using allocsize to setting 1gb allocsize. Also GB/s class filesystem. Cheers, Ivan Novick |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH 1/2] xfs: dynamic speculative EOF preallocation, Dave Chinner |
|---|---|
| Next by Date: | Re: allocsize mount option, was: [PATCH 1/2] xfs: dynamic speculative EOF preallocation, Michael Monnerie |
| Previous by Thread: | Re: [RFC, PATCH 0/2] xfs: dynamic speculative preallocation for delalloc, Dave Chinner |
| Next by Thread: | Re: [RFC, PATCH 0/2] xfs: dynamic speculative preallocation for delalloc, Michael Monnerie |
| Indexes: | [Date] [Thread] [Top] [All Lists] |