[Zope-DB] RFC: mxODBCZopeDA & ZEO

[Chris Withers]

Charlie Clark wrote:

>>Myself? I setup a whole instance in Subversion and check it out onto any 
>>machien
>>I want a client one...
> 
> Interesting. I assume this involves using some kind of file storage?

Nope, I set up an "empty" instance with no Data.fs. The file storages are not 
kept in the Subversion repository...

> Yes, wasn't possible in the test configuration but maybe useful for support 
> enquiries.

Indeed. What was so special about the test configuration?

> From website to cluster.

Cluster of what?!

> The current state sets up identical copies of the 
> connection object in the various ZEO clients. They have to communicate 
> through a local ODBC manager with the database. In a sense this goes against 
> something like the mxODBCZopeDA which should be able to handle all necessary 
> connections from all ZEO clients in a single object. This looks like it's 
> going to remain a theoretical difference as there is unlikely to be any 
> performance difference either way.

Exactly ;-)

> We're essentially looking at the question: what does ZEO give you and how 
> does it work?

ZEO is for ZODB. It brings prettymuch nothing if you just use DA's.
If you use ZODB storages for data (FileStorage, APE, OracleStorage - bwahahaha) 
then it buys your very easy clustering...

> However, it looks like the common setup avoids the pitfalls of separate 
> physical installations.

Um?

> That's good to hear. There doesn't seem to be much running with the 
> mxODBCZopeDA which seems to suggest that such sites primarily use the ZODB 
> for storage.

Indeed. ZEO is important if you use ZODB for storage or if rendering of your 
site takes significant horsepower.

cheers,

Chris

-- 
Simplistix - Content Management, Zope & Python Consulting
            - http://www.simplistix.co.uk