[Ext2-devel] Re: [RFC] mke2fs with DIR_INDEX, RESIZE_INODE by default

[Arjan van de Ven]

On Sun, 2006-03-19 at 21:24 -0500, Theodore Ts'o wrote:
> On Sun, Mar 19, 2006 at 10:59:26AM +0100, Arjan van de Ven wrote:
> > 
> > > In general, I doubt directory readahead actually buys you *that* much,
> > > because most workloads follow up the readdir with a stat() or an
> > > open() call for each file returned, or something which requires
> > > reading in the inode.  In addition, it's rare that the directory will
> > > be contiguously allocated, which also cuts down on the value of the
> > > readahead.
> > 
> > what it buys you shows up on raid1 I suppose... while disk 0 is doing
> > the head seek for the stat/open, disk1 is doing the head seek for the
> > next directory entry.... so I can see that being a big gain in that
> > specific circumstance.
> > 
> 
> As long as you have the spare disk bandwifth and disk1 isn't needed
> reading the directory entry for some other make process as part of the
> "make -j16" kernel build...

sure.... on the flipside.. if the readahead request is in the elevator
it may be merged with that make ;)
so I agree, it's not even a guaranteed win on raid1