As an experiment, I started porting zodb to amazon s3 storage a while ago...

You can use it to start zope, but it makes far too many writes inefficiently (it took about 15 minutes for zope2 to create all it's objects on a first start).

Now that ec2 comes with static IP addresses, I might end up looking at it again. It needs refactoring to write only one file per transaction (or savepoint), but once done I would imagine it might then offer reasonable performance.


Tim Nash wrote:
The google summer of code is almost underway and already has a lot of
good ideas but I was wondering if any of my fellow zope users would
have any interest in the following area:

I don't have the skills but I think it would be cool if some
student ported Zope to utilize features of the google file system or
libferris. Libferris is a virtual file system that mounts just about
everything including postgres, xml and OpenOffice docs.  If zope ran
on the gfs (primarily adding business logic, security and publishing)
it would give a boost to the value of any zope based company. Also, I
think it would be fun to run map/reduce on my stored objects! A zope
that ran on libferris and offered secure publishing of libferris
resources would just be cool.

If anybody else thinks this would be useful I will jon the gsoc user
list, (even though I'm not a student and I'm not experienced enough
with zope to be a mentor) and do my best to make this happen.
Zope maillist  -
**   No cross posts or HTML encoding!  **
(Related lists - )

Zope maillist  -
**   No cross posts or HTML encoding!  **
(Related lists - )

Reply via email to