Return to BSD News archive
#! rnews 6952 bsd
Message-ID: <333C1614.ABD@sgi01.grn.aera.com>
Date: Fri, 28 Mar 1997 12:03:48 -0700
From: Lee Ward <lee@sgi01.grn.aera.com>
X-Mailer: Mozilla 3.01 (X11; I; IRIX 6.2 IP22)
MIME-Version: 1.0
Newsgroups: comp.unix.bsd.freebsd.misc,comp.unix.bsd.bsdi.misc,comp.sys.sgi.misc
Subject: Re: no such thing as a "general user community"
References: <331BB7DD.28EC@net5.net> <5g9hjp$api@flea.best.net> <5gmb58$6jd$1@news.clinet.fi> <5gn3ig$83d@flea.best.net> <5goqrq$5ak$1@news.clinet.fi> <5hd29s$e7t@fido.asd.sgi.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 207.40.197.84
Lines: 128
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!cpk-news-hub1.bbnplanet.com!su-news-hub1.bbnplanet.com!news.bbnplanet.com!news.sprintlink.net!news-peer.sprintlink.net!news-pull.sprintlink.net!news.sprintlink.net!news-pen-14.sprintlink.net!superego.idcomm.com!207.40.197.84
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:38011 comp.unix.bsd.bsdi.misc:6497 comp.sys.sgi.misc:29510
Larry McVoy wrote:
>
> Mika Ruohotie (mickey@cantina.clinet.fi) wrote:
> : what is the major difference between xfs and ufs? i am still under impression
> : that on a news server environment xfs would perform more efficienty. yet i
> : admit that the latest development on filesystems on freebsd have made some
> : serious leaps. and the gap is probably narrow as it's been pointed out.
>
> XFS is a journaling file system. Do an untar and walk up to the SGI and
> turn the power off. Do the same thing to your FreeBSD machine. Then time
> the boot.
>
This, indeed, is a very nice feature of XFS.
> XFS includes striping and mirroring. Striped XFS file systems move data
> at 500MB/sec. They have done so for years.
>
> XFS is integrated with SGI's NFS. SGI's NFS has an extension, a freely
> available under the GPL extension, that delivers 85MB/sec read or write
> rate, over the wire, 1 process, single threaded, no async I/O. In other
> words for (whatever) { read(fd, buf, 1<<20); }.
>
These numbers seem pretty wild to me. How many hundreds of thousands of
dollars worth of hardware are we talking about here? I've been working
with super-computers on and off for the last few years - many SGI
machines included. I've never seen an SGI deliver even the above claimed
NFS rate on a local file system. While it seems that it *could* be done,
I've just never seen it. I realize this is only anecdotal but it seems,
to me, as valid as the original unsupported statement.
> : how far from having working lfs we are, or does it work already? i think
> : last time i asked about xfs i was told lfs "is getting there" but wasnt
> : quite working...
>
> LFS is a joke. Check out the old Usenix articles on it, there is one where
> I made Berkeley publish what essentially amounts to a retraction of their
> claims.
>
> : also, if someone could provide me some stats how many files/sec xfs can
> : write to disk compared to ufs, since i believe that's one of the ways to
> : compare those two, xfs were, last time i heard, able to overperform ufs.
>
> Sure can.
>
> File & VM system latencies in microseconds - smaller is better
> --------------------------------------------------------------
> Host OS 0K File 10K File Mmap Prot Page
> Create Delete Create Delete Latency Fault Fault
> --------- ------------- ------ ------ ------ ------ ------- ----- -----
> P5-133 FreeBSD 2.2-C 2857 1123 4000 3225 133 -1 0.0K
> R10K IRIX64 6.2-no 374 389 353 387 55 -1 21.5K
>
> About a factor of 10. The R10K is 2x the perf of the P5 but the FreeBSD
> system is sitting idle waiting for the desk. An infinitely fast CPU
> wouldn't help.
>
Those do seem to be significant in this argument. For me, I'm usually
interested in I/O performance to and from a file. Not so interested in
operations in the name space. This seems to be more the norm for
scientific computing than how fast a file can be created or deleted.
Since I hadn't run any sort of a benchmark in awhile, I compiled Bonnie
and ran it just for a quick idea.
Two machines, an SGI Indy R5000 (180 Mhz), IRIX 6.2, with 64MB of memory
and the delivered system disk (reported as "SGI IBM DORS-32160
WA6A"). The other machine was a P5-100, BSDI 3.0, with 16 MB of memory.
The disk controller reports as "BusLogic BT-946C rev 4.25J (32-bit)".
The attached system disk reports as "SEAGATE ST32430N rev 0300 (SCSI-2)"
For the Indy, I got:
-------Sequential Output-------- ---Sequential Input--
--Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block---
--Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
/sec %CPU
100 2894 68.2 4170 25.0 1665 16.5 2240 52.3 4595 24.0
37.9 4.0
For the PC, I got:
-------Sequential Output-------- ---Sequential Input--
--Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block---
--Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
/sec %CPU
100 1579 37.1 4641 28.6 632 6.1 1016 23.9 3695 19.4
15.6 1.8
As I would have expected, the character based performance was much
better on the Indy. I'm not as impressed with the block oriented I/O as
the Indy was far more than twice the price of the PC.
> Don't get me wrong. I think free software is great. I think Matt is
> one smart cookie, I'd love to hire him. Ditto for the FreeBSD team.
> But make no mistake there is more to an OS than a simple minded file
> system or a context switch. SGI's OS scales to 100s of CPUs, supports
I'd have to agree with that statement. However, the most expensive tool
isn't always the best for the job is it? I mean, Ferrari's cost orders
of magnitude more than most pickup trucks. They're not very good at
pulling stumps though. Similarly, for a small group taking a small news
feed, it may be approrpiate to use a lower cost PC than any of the SGI
offerings. Then again, it may not. It seems to me that it would be
useful to weigh the benefits and costs.
> more devices, and has more stuff that people with $$$ want. This is
> serious software with serious people working on it. Do you want IRIX for
> a desktop to read news? Hell no, install Linux or FreeBSD or whatever
> on your 386 and you are done. If you have real work to do you may find
> that you have to pay for your tools.
What is "real" work anyway?
Now then, don't get *me* wrong. I'm not trying to bash either of these
operating systems. Each has it's own place. It's just that your
implication that the BSD operating systems are toys that don't have
"serious software" being used by "serious people" is degrading.
> --
> ---
> Larry McVoy lm@sgi.com http://reality.sgi.com/lm (415) 933-1804