[ZODB-Dev] ZEO bandwidth requirements?

[Jim Fulton]

"Barry A. Warsaw" wrote:
> 
> >>>>> "JF" == Jim Fulton <jim@zope.com> writes:
> 
>     JF> Or make the BerkelyStorage uncompress the pickles when getting
>     JF> the references (it would still store compressed data, which
>     JF> might help us on Berkeley cache requirements, unless it
>     JF> doesn't). Given the large cache that the Berkeley storage
>     JF> needs, I'm wary of adding another table to the mix.
> 
> That's a tradeoff, because sniffing the pickles (along with u64/p64
> conversion, which Jeremy improved) accounts for about 1/6 - 1/5 of the
> time for _finish().  If we add uncompressing before pickle sniffing, I
> imagine that ratio will go up. :)

Wow! I'm surprised it's that much of the time. Are you sure about that?
Is this in the best-case performance case?

It's tempting to rearrange the pickle format to make this info easier
to get at.  For example, it might be possible to get the persistent 
reference info outside the pickle so that it can be handled independently.
Hm....

Jim

--
Jim Fulton           mailto:jim@zope.com       Python Powered!        
CTO                  (888) 344-4332            http://www.python.org  
Zope Corporation     http://www.zope.com       http://www.zope.org