Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 3 Jan 2016 13:20:38 -0600 (CST)
From:      Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
To:        "Mikhail T." <mi+thun@aldan.algebra.com>
Cc:        freebsd-fs@freebsd.org
Subject:   Re: NFS reads vs. writes
Message-ID:  <alpine.GSO.2.01.1601031314540.28454@freddy.simplesystems.org>
In-Reply-To: <56895E2E.8060405@aldan.algebra.com>
References:  <568880D3.3010402@aldan.algebra.com> <alpine.GSO.2.01.1601031006020.28454@freddy.simplesystems.org> <56895E2E.8060405@aldan.algebra.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 3 Jan 2016, Mikhail T. wrote:
> 
> Why would the same ZFS -- with the same slog -- be working faster, when written to locally, than when over NFS?

Yes (it depends).  Normal local writes are async writes so they do not 
use the slog at all.  NFS writes are usually synchronous writes so 
they hit the slog hard.

You would need software which intentionally issues synchronous writes 
to local disks in order to test synchronous local writes.

Write sizes of 128k and above are written directly to pool disks and 
don't go to the slog, even if they are sync writes.  This is done to 
make sure that the slog does not limit pool performance due to 
bandwidth limitations (vs IOPS).

Bob
-- 
Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.GSO.2.01.1601031314540.28454>