[Zope-dev] NEO High Performance Distributed Fault Tolerant ZODB Storage

[Vincent Pelletier]

Le mercredi 31 mars 2010 18:32:31, vous avez écrit :
> A few questions that you may want to add in a FAQ.

We started that page and will publish it very soon, based on most points you 
raised. Other pages are also being worked on, such as an overview of a simple 
NEO cluster.

>   - Why not include ZODB 3.10/Python 2.6 as a goal of the project?
>     - I understand *today* the technologies use python 2.4 but
>       ZODB 3.10/Plone 4/Zope 2.12 use python 2.6

We indeed aim at supporting more recent versions of python and Zope.

Actually, your remark made us realise that our functional tests are currently 
(accidentally) running in a mixed 2.4/2.5 python environment: test process is 
started explicitly with 2.4, and forked processes (masters, storage and admin 
nodes) are running on default python, which is 2.5 (as of Debian stable).

The standard Zope version in Nexedi is 2.8, which explains why we want to 
support it. We will switch to 2.12, as we have ERP5 unit tests running on 2.12 
for some weeks[1] now. NEO will move to 2.12 at the same time or earlier.

>   - Maybe explain the goal of the project clearer:
>     """
>     NEO provides distributed, redundant and transactional storage designed
>     for petabytes of persistent (python?) objects.
>     """

Thanks, updated.

>   - A buildout for NEO would lower bar for evaluation

This is on our roadmap (...to be published along with the FAQ), but the 
priority currently goes to 2 developments which might/will break 
compatibility: pack support (required an undo rework which was recently 
integrated, pack itself needs more unit testing prior to integration) and 
multi-export support (aka "ZODB mountpoints", also in need for more testing 
before integration).

[1] http://mail.nexedi.com/pipermail/erp5-report/ ("_z212" in subject)
-- 
Vincent Pelletier