I believe, after my discussions with the folks at Transarc, that the
writeup claims that network bandwidth, as well as file server throughput
would be adversely affected if AFS kept track of the atime field. I
believe there was some belief that the ctime field might be accomodated in
the future if enough people requested it. I put us down for being in favor
of a ctime field.
-Brian
On Mar 2, 3:45pm, Robert Banz wrote:
} Subject: Re: File access/modification times
} On Thu, 25 Feb 1999, Brian Buhrow wrote:
}
} > AFS doesn't store access times. The mtime, ctime and atime fields of
} > the stat structure are all filled in with the mtime. Any programs that
} > rely on the difference between access and modification times are going to
} > behave strangely when working on files in AFS. I believe there is an
} > official writeup from Transarc on this.
}
} This line of thought just brought up a question that's been in my head for
} awhile -- Some OS's have an option on mounting their native filesystem
} that turns off 'atime' updates. Does AFS use the native filesystem's
} 'atime' for anything in particular? I would assume that one would remove
} a bit of a load from a filesystem by needing to update atime's on read
} operations, and henseget a little performance boost...
}
} -rob
}
}
>-- End of excerpt from Robert Banz
