Coda File System

Re: Few Basic Coda questions...

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Fri, 28 May 1999 09:24:33 -0400 (EDT)
Hi, 

> 
> =====>" Browsing the network with huge number of binary files may be a
> common activity of windows users. User may just browse the directory to
> look for some information, but never execute any of those
> programs...bringing down those large binaries may really slow down the
> performance without any gain ( I mean no cache hit in near future ). 
> 
> Is there any interest to solve this problem? Any ideas?"

One could add block level caching.   It's not that complicated - the
kernel already knows about sparse files, so if a block isn't found in our
cache file you fetch the surrounding 64k.  

In a mobile environment this can lead to disappointment, e.g. you might
miss a block of WP.

> 
> ===> "I understand after the first time cache by Venus next access to
> files will be faster compare to any other protocol hitting wire. But I
> don't understand how its faster compare to local disk browsing. Is there
> something I am missing?"
> 

Coda has a very sophisticated directory lookup cache, which is faster than
the Windows lookup operations for FAT.  We are talking about very small
differences
though.

- Peter -
Received on 1999-05-28 09:26:01