[ZODB-Dev] ZODB blob support.

[Toby Dickenson]

On Thursday 17 March 2005 19:59, Chris McDonough wrote:
> Hi folks,
> 
> In preparation for some work that will tentatively happen at the Pycon
> sprints, I've put up a very basic proposal outlining "blob" support for
> ZODB.  It would help if people with a stake in ZODB had the time to
> comment on it before work starts (please comment here, as opposed to on
> the proposal itself, just so we get the widest audience).
> 
> http://www.plope.com/Members/chrism/zodb_blob_proposal

Replying here to some comments in that wiki.....

| Prevent blob data from evicting "more important" data that lives in
| a ZODB or ZEO cache. 

Rather than having blobs bypassing the normal caches, I think the right 
solution is to have these caches using a smarter cache eviction policy. 

The connection cache uses least-recently-used, and the ZEO cache uses an 
approximation of lru. A better cache eviction algorithm would be aware of 
different object characteristics. IMHO blobs are only one not-so-special 
case. I suspect there would also be a benefit to making those tiny internal 
BTree nodes particularly cache-sticky. Ive written about this a little 
before: http://mail.zope.org/pipermail/zope-dev/2003-April/019418.html

| some benchmarks I ran a while ago indicated that the current Pdata
| solution for OFS.Image suffers greatly when used with ZEO

Thats not only true for blobs surely. I guess some prefetching/readahead would 
help alot of ZEO benchmarks.

-- 
Toby Dickenson