Re: Please help me tune FreeBSD for bulk.fefe.de/scalability/

From: Sean Chittenden (sean_at_chittenden.org)
Date: 10/25/03

  • Next message: Devon H. O'Dell: "Re: Please help me tune FreeBSD for bulk.fefe.de/scalability/" 
    Date: Fri, 24 Oct 2003 15:27:43 -0700
To: Felix von Leitner <felix-benchmark@fefe.de>

    > Please also tell me if you have suggestions for other benchmarks
    > that say something about the scalability of an operating system,
    > that I could (and should) include in my benchmark suite.

    >From a database perspective, I have two benchmarks that I'd be
    interested in seeing.

    1) Time required to read a frequently accessed page that has been
       mmap()'ed. As you state, it takes FreeBSD longer to mmap() the
       first page, but I'd like to see the benefit for the startup time
       documented in a benchmark. Presumably, the extra time required to
       load the first page is worth something.

    2) Time required to read a frequently accessed page that has not been
       mmap()'ed (ie, raw read() IO). I'd like to know how fast each OS's
       file system cache is.

    A few other comments about your benchmarking (which are really great
    to have, thank you!):

    1) Are you using the most recent version of FreeBSD-CURRENT? If so,
       I'd be curious to have libpthreads/libkse benchmarked. Time to
       spawn a new thread and time required to join ten threads.
       Something like that would be quasi-useful.

    2) You state in your mmap benchmark section:

       "As you can see, Linux 2.4 appears to scale O(n), while Linux 2.6
       is O(1). FreeBSD looks to be much faster than Linux 2.6, but you
       need to keep in mind that FreeBSD took an extraordinary time to do
       the actual mmap, so this good result does not save the day for
       FreeBSD."

       I would disagree in that it is the common case for an mmap()'ed
       page to be simultaneously mmap()'ed by other processes (think
       Apache serving a busy homepage: each proc is mmap()'ing the same
       file). I'd imagine that FreeBSD would win in a performance test of
       this nature because of the work it did when it mmap()'ed the first
       page, therefore think that this proves that FreeBSD is justified in
       the extra time that it takes when mmap()'ing a new page.

    3) FreeBSD isn't finished in terms of its locking. That said, these
       benchmarks really impress me that even though many subsystems are
       still under the Giant lock, FreeBSD 5.X performs as well, if not
       better than Linux 2.6.

    -sc 

    -- 
Sean Chittenden
_______________________________________________
freebsd-performance@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-performance
To unsubscribe, send any mail to "freebsd-performance-unsubscribe@freebsd.org"
  • Next message: Devon H. O'Dell: "Re: Please help me tune FreeBSD for bulk.fefe.de/scalability/"

    Relevant Pages

    • Re: Linux users seeks some freeBSD pointers (Benchmarks)
      ... Or a decent database benchmark, ... example do the FreeBSD people as a whole, really give a shit about how ... It's the sort ... > sort of personal axe to grind to try and tell me otherwise. ...
      (comp.unix.bsd.freebsd.misc)
    • Re: Linux users seeks some freeBSD pointers (Benchmarks)
      ... Yes FreeBSD performs good but your statement ... ... Look at the first graph. ... In the write-up on his first benchmark he demonstrates his bias and he ... realising they're all largely worthless? ...
      (comp.unix.bsd.freebsd.misc)
    • Re: (mySQL) benchmarks strike back
      ... > advocacy for FreeBSD, so I'm sure my boss is going to ... Perhaps Linux has gained ... benchmark test results should not be the ... only criteria for selecting an operating system. ...
      (freebsd-questions)
    • Re: Benchmark: NetBSD 2.0 beats FreeBSD 5.3
      ... > the benchmark you cited is for uniprocessor systems only. ... > It's comparing apples with oranges. ... No, many users, me included, only run FreeBSD on UP systems. ...
      (freebsd-hackers)
    • a possible explanation for the mmap benchmarks
      ... Since it does mmap sequentialy, free space linear search vs hint ... splay tree degenerates to a list. ... The questions is whether altered benchmark, ...
      (freebsd-hackers)