Re: Can O_SYNC be implemented by using fsync?

From: Chris Wedgwood (cw@f00f.org)
Date: Mon May 15 2000 - 15:59:07 EST


[...]

    (i.e. on NetWare, Oracle uses the DirectFS interface which by-passed the
    cache completely). Oracle also provided it's own caching above the OS
    for remote SQL users. When they called DirectFS, it was always with an
    "O_SYNC" semantic, which I think is the spirit here.

Linux right now how no way to bypass the buffer-cache when writing to
a filesystem.

O_SYNC is as good as you can get, you want to be sure you most
recently issued write(...) has hit the platters (polluting the
page-cache whilst undesirable, is unavoidable at present).

    I think this is probably OK.

>From profiling Oracle under Slowaris, I would say Oracle _always_
writes expecting the data to hit the platters immediately (I could be
wrong).

Because linux does not yet support this I assume they use O_SYNC,
hence it must be as fast as possible or things like Oracle will
really suck.

  --cw

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon May 15 2000 - 21:00:26 EST