Infrastructure

The Wide Finder host is at wf2.network.com

  • Sun Fire T2000
  • CPUs: UltraSparcT1, 32 cores
  • Disks: 73GBx2 (mirrored as system disk), 136GB x 2
  • RAM: 32GB
  • OS: Solaris 10 U4
  • Patches: Solaris 10 Recommended patches 03/29/08 release.

If you want an account, first you will need to implement a program that produces correct results on The Benchmark using the 100,000-line sample. Email me at tim dot bray at sun dot com, explaining your Wide Finder interest, and I'll send a pointer to the sample data.

The pkg-get command can be used to install blastwave software, see http://www.blastwave.org/packages.php, but you have to be root. I've also made /usr/local world-writeable, so anyone with an account should be able to compile & install pretty well anything from source code. See the following links for usage notes on installed software:

If you get an account, you probably want to look at ~twbray/.bash_profile to figure out the archaic Solaris $PATH voodoo. It's a pity Indiana isn't running on SPARC yet, it's got that friendly familiar GNU flavor.

Interesting filesystems:

  • /export/home - home directories, 27G
  • /wf1 - ZFS pool, 286G
  • /wf1/data - ZFS filesystem, all options defaulted.
  • /wf1/data/logs - logfile test data
  • * O.all - Apache httpd log data, 45,275,432,643 bytes in 218,201,129 lines
  • * O.1k - 1,000-line sample of O.all
  • * O.10K - 10,000-line sample of O.all
  • * O.100K - 100,000-line sample of O.all
  • * O.1m - 1,000,000-line sample of O.all
  • /wf1/data/compressed - individual compressed logfiles used to create O.all

To get a feel for the underlying I/O performance, here are the results of a 100G Bonnie run:

              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    GB M/sec %CPU M/sec %CPU M/sec %CPU M/sec %CPU M/sec %CPU  /sec %CPU
T2K       100  20.5  100 150.1  100  61.4 64.8  19.8 98.9 148.9 76.7   214 10.7

That CPU percentage is percent of one core. The results show that while the I/O performance is pretty respectable (about 150M/sec in and out), you can easily saturate one of those wimpy little cores; for example, I'm pretty sure the ~20M/sec sequential-I/O numbers are CPU-limited.

Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.
  1. May 05, 2008

    no_one says:

    Tim - is atime off on all those zfs filesystems? If not, can you turn it off a...

    Tim - is atime off on all those zfs filesystems?
    If not, can you turn it off and re-do the bonnie run?

    Something like:
    zfs set atime=off wf1
    zfs set atime=off wf1/data
    zfs set atime=off wf1/data/logs
    etc

    Is compression off on wf1 and wf1/* ?

    turning off exec is probably a good idea, too - unless people are moving their compiled app to the data folder before running them or something.

    http://docs.sun.com/app/docs/doc/817-2271/gayns?a=view

  2. May 05, 2008

    ErikEngbrecht says:

    Is Java installed? What version? re: /usr/local Are you sure it is wise to ha...

    Is Java installed? What version?

    re: /usr/local

    Are you sure it is wise to have everyone installing into /usr/local? What if different people need to run different versions of the same package? What if people break each other's configuration?

Sign up or Log in to add a comment or watch this page.


The individuals who post here are part of the extended Sun Microsystems community and they might not be employed or in any way formally affiliated with Sun Microsystems. The opinions expressed here are their own, are not necessarily reviewed in advance by anyone but the individual authors, and neither Sun nor any other party necessarily agrees with them.

© 2010, Oracle Corporation and/or its affiliates
Powered by Atlassian Confluence
Oracle Social Media Participation Policy Privacy Policy Terms of Use Trademarks Site Map Employment Investor Relations Contact