POOP / Python (Object Oriented Persistence) ?

Martijn Faassen m.faassen at vet.uu.nl
Mon Jul 10 11:43:12 EDT 2000


"Tony J Ibbs (Tibs)" <tony at lsl.co.uk> wrote:
> Martijn Faassen explained that Zope does indeed do most of the things I'd
> mentioned, and then:
>> > persistence over LARGE networks, etc.
>>
>> What would this mean?

> One of the interests of the people in Glasgow is how one copes with data
> that may be distributed over many computers. This includes issues such as
> how do you fallback to another copy of the data, how do you decide
> whether/when to cache locally, and if so when to update, etc.

Zope's ZEO is going into this area. I haven't looked at it myself, but
basically you can cluster several copies of the ZODB together. In the
case of ZEO there's the case of a single point of failure (the central
front-end machine), however. But ZEO is definitely very cool. Not a lot
of code either from what I've seen.

> Since they
> also store the program code in the store, this also means sharing software
> over the network as well.

Right, I don't think ZEO has looked into this yet. You'd have to have your
Zope products (or in case of raw use of the ZODB, python classes) installed
on all servers for it to work. Since Zope is fairly componentized and
part of a Zope Web App tends to be in DTML and is in the ZODB already 
(with more Python moving into the ZODB too), this doesn't seem to be
a large project just yet.

> In todays emphasis on shared services over the
> net, a lot of this begins to sound remarkably prescient.

Definitely! When did this research take place?

> It's often regarded as one of those "we'll worry about that later" problems,
> since people expect to be working on a local store, but thinking about it
> earlier can have interesting insights into what one should be *doing*
> locally (and as soon as one has two users on two systems sharing stuff, the
> slippery slope has started). But it did seem to be a *difficult* problem,
> conceptually as well as practically.

Agreed that it's important, agreed that it's difficult. Luckily Zope has
Jim Fulton. :) Again, ZEO is a remarkably small amount of code.

> But I *know* I don't understand the field, so I'll shut up straight away
> about it...

I don't understand the field either, so let's babble on. Perhaps we'll
learn something. :)

>> Do you still think what the ZODB does is much
>> different from what you're describing?

> Erm, no, it sounds like I need to find some excuse to find out more about
> Zope, instead! (damn, another worthwhile thing to learn about).

This article by Andrew Kuchling may be interesting, if you're just interested
in the ZODB and ZEO by itself:

http://starship.python.net/crew/amk/python/writing/zodb-zeo.html

Regards,

Martijn
-- 
History of the 20th Century: WW1, WW2, WW3?
No, WWW -- Could we be going in the right direction?



More information about the Python-list mailing list