Date: Thu, 17 Feb 2000 12:49:04 -0800 From: Alfred Perlstein <bright@wintelcom.net> To: Brad Knowles <blk@skynet.be> Cc: Parag Patel <parag@cgt.com>, current@FreeBSD.ORG Subject: Re: FWIW: More questionable softupdates+vinum benchmarks Message-ID: <20000217124904.S3509@fw.wintelcom.net> In-Reply-To: <v04220829b4d205529a91@[195.238.1.121]>; from blk@skynet.be on Thu, Feb 17, 2000 at 09:14:20PM %2B0100 References: <74982.950812517@pinhead.parag.codegen.com> <v04220829b4d205529a91@[195.238.1.121]>
next in thread | previous in thread | raw e-mail | index | archive | help
* Brad Knowles <blk@skynet.be> [000217 12:45] wrote:
> At 10:35 AM -0800 2000/2/17, Parag Patel wrote:
>
> > Hello. I have a friend's quad PPro box temporarily sitting in my garage
> > that I've been using to play with 4.0-CURRENT and vinum. Since the last
> > series of bug-fixes a few weeks ago, everything works as advertised.
>
> While we're on this subject, I recently did some benchmarking
> with just a single disk on a machine running 3.4-STABLE, both with
> and without softupdates. I haven't yet gotten a chance to test it
> with vinum and softupdates, but what I got did a pretty good job of
> impressing me.
>
> I won't quote the mail messages here (you can go read
> freebsd-stable for that, look for message-ids
> <v0422080eb4d1975ec780@[195.238.1.121]> and
> <v04220816b4d1a39aa381@[195.238.1.121]>), but suffice it to say that
> for my testing, I saw 3-5x performance increase for the benchmark I
> was running ("postmark", from Network Appliance).
>
>
> The primary reason I was doing this was to get some sort of idea
> what kind of performance you might be able to expect if you took an
> application like a mail server or a USENET news server (with
> traditional spool), and you re-wrote the application so as to avoid
> expensive synchronous meta-data operations (which are the things that
> softupdates tries to optimize out of existance).
>
> I'm starting to actually believe that you might really be able to
> see a real-world 5-10x performance increase, if you did that. In
> fact, I believe that this is pretty much what admins of boxes running
> INN are seeing when they use CNFS. Jeez, Louise.
Depending on how temporary your temporary files are, it'd be
interesting to see if the 4.0 optimizations benifit your benchmark
and also remain stable.
Would it be possible for you to grab a 4.0RC and let us know?
thanks,
-Alfred
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000217124904.S3509>