On Tue, 24 Oct 2006, Joel VanderWerf wrote:
>> The speed of the approach is tied very directly to the speed of the IO 
>> system, but it's relatively fast and very portable so long as your file 
>> naming scheme is itself portable.
>
> Kirk,
>
> Did you get a feeling for what the break-even point was? In other words,
> how many files do you need before the depth 2 scheme beats a single flat
> dir?
>
> Playing around with this idea in fsdb, with 100_000 files, access times seem 
> to get worse as depth varies from 0 to 2. Creation time was better at 
> depth==1 than in either of the other cases.

Not really.  The IO performance on my test machine is so bad that it 
didn't seem to make a significant difference.

You have me curious now, though, so I may go test it on a box with better 
IO and see what I can find out.


Kirk Haines