Date: Sun, 28 Jan 1996 14:26:43 -0800 From: "Russell L. Carter" <rcarter@geli.com> To: John Brann <jbrann@panix.com> Cc: hackers@freefall.freebsd.org Subject: Re: Good news -- pipe stuff Message-ID: <199601282226.OAA13790@geli.clusternet> In-Reply-To: Your message of "Sun, 28 Jan 1996 14:21:47 EST." <199601281923.OAA07191@jbrann>
next in thread | previous in thread | raw e-mail | index | archive | help
} owner-hackers-digest@freefall.freebsd.org wrote...
} > ----------------------------------------------------------------------
} >
} > From: "Jordan K. Hubbard" <jkh@time.cdrom.com>
} > Date: Sun, 28 Jan 1996 01:21:01 -0800
} > Subject: Re: Good news -- pipe stuff
} >
} > > I read the Lai/Baker paper: in "Figure 1", FreeBSD kicks some serious
} > > butt on context switches -- it appears to be both flat and linear past
} > > 200 processes (the limit of the graph in the figure).
} >
} > Actually, at a later talk it came out that Linux had substantially
} > improved this in the current release. However, both this talk and
} > Larry McVoy's talk that followed it were wastes of paper and time for
} > all concerned.
} >
} [remainder of Jordan's analysis removed]
}
[stuff deleted]
} I downloaded and read the Lai/Baker paper. I agree that it was clearly
} shallow, but I don't think I've ever seen any worthwhile cross-OS or
} cross-machine performance comparisons. Anything beyond the grossest
} conclusions (Solaris seems a bit slow) are hard to be convincing about.
}
} What I read into the paper was;
}
} 1. Linux seems very fast as a standalone system. Much (more) kudos
^^^^^^^^^^^^^^^^ hmmm.
} to Linus and his followers.
}
} 2. FreeBSD is also very fast, with some suggestion that there might
} be some things in the filesystem that would benefit from looking
} at.
}
} 3. Solaris bites. (This is influenced by personal prejudice, I
} run it at work).
}
} Something about this clearly got under Jordan's skin, and for the first
} time (and I do always try to read as much of his postings / e-mails as
} I can) I detected real irritation in his analysis. I'm disappointed in
} this, it jarrs with all those "do not get into that Linux vs. FreeBSD
} argument" articles of his.
Er, but there is this little piece of Unix functionality called "networking".
Solaris, FreeBSD, and just about every Unix sold today does it fine.
Important parts of Linux networking are lousy. Unusable, even,
contrary to McVoy's claims.
The contortions used to cover this up, in this article, and for instance
in a Usenet posting by Larry where he quotes beta SGI numbers for FDDI to
indicate the certainty of a RSN Linux NFS fix irritates *me*
occasionally.
If Stanford University needs high performance networking testbeds, they
should contact Geli Engineering Labs :-) to see about donations of time
on systems that will kick some serious networking butt.
}
} For what it's worth, here are my conclusions after almost a year following
} FreeBSD's fortunes.
}
} 1. FreeBSD is _behind_ Linux in support for many useful features -
} devices, threads, ELF and, no doubt, other things.
}
} 2. FreeBSD is _ahead_ of Linux in focus. For good or ill (and I
} think it's good) FreeBSD has a core team which, collectively,
} with leadership provided by the normally saintly Jordan,
} guides the system forward in a coherent fashion. I don't watch
} Linux closely, but the entry of commercialization will probably
} fragment the it even further than it has already gone.
Oops, I need to add to this list:
3. FreeBSD is _ahead_ of Linux in networking. In fact, it is
_ahead_ of Solarix x86 in that regard, if Fast Ethernet
is of any interest.
}
} It's common to read in our mailing lists of people converting from Linux to
} FreeBSD. I don't know if the reverse can be read in the Linux groups and
} lists, but in any case, FreeBSD must be doing _something_ right. I
^^^^^^^^^^^
Now what on earth could that be? :-) :-)
Cheers,
Russell
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199601282226.OAA13790>