[Zope3-dev] Re: [ZODB-Dev] ZODB4 project plan

[Nick Pavlica]

Jeremy,
  The items I listed in my wish list were based on a handful of features =
that=20
I would like to have in any Enterprise level database.  In the company th=
at I=20
work for, we have projects that deal with data sets as small as a few MB,=
 and=20
some up to 30TB/day.  We use different tools to best match the needs of e=
ach=20
project, but there are some core database features that would make life=20
easier in general.  Having good general purpose databases allow us to use=
=20
fewer tools (databases) to get our jobs done.  We are slowly moving in th=
e=20
direction of OO databases, because we are using OO tools like Java, C++, =
and=20
some Python, which makes the development of new apps cleaner than using t=
he=20
layers of SQL, JDBC, object relational mapping tools, etc.  Currently we =
are=20
using the GOODS OODB on one of our larger projects.  Because Python has b=
oth=20
productivity, and Quality in mind it would be nice to have a database lik=
e=20
ZODB that works very well with Python, but can fill the requirements of=20
critical applications.  Postgres, and MySQL are doing a good job of tackl=
ing=20
SQL world, but I think that there is allot of opportunity for an OODB to =
make=20
a name for itself in the OpenSource community.  I believe that Digital=20
Creations could broaden their opportunities by looking at the ZOPE=20
application server, and the ZODB as two separate business models(training=
,=20
books, consulting,...).  I'm sure you all know this, but I feel that my=20
opinion counts, at least a little bit :).  I truly believe in the OpenSou=
rce=20
technology world and try to contribute to it as much as I can.  I would b=
e=20
willing to help with this projects as well.

> Many of the features you mention are already supported in ZODB3.
-- Where can I find a "Comprehensive"  feature list ? =20
-- Is there a white paper for ZODB3?=20

>  Can  you be specific about what is lacking for "distributed" to pick t=
he
> first example?
--  I realize that "Distributed" can mean different things to different p=
eople=20
and is rather vague and often over used in the DB world.  Distributed in=20
terms of what I was referring to, is the distribution of a single Databas=
e=20
across multiple physical servers/storages.  For example, we are developin=
g an=20
application that will be used in various physical locations with a server=
 at=20
each of them.  However, the client application will not be aware that it =
is=20
using objects that are physically stored half way across the country.  It=
 is=20
just concerned with the logical database.  This is nice because it allows=
 us=20
to keep the clients simple, while being able to focus and refine the=20
database.  I added some links to a respected publication on Distributed=20
databases, that may be of interest to you or others in this group.

Distributed DB Book:
http://www.cs.ualberta.ca/~database/ddbook.html

Distributed DB notes:
http://www.cs.ualberta.ca/~database/ddbook/notes.html

> Also, I'll have to comment on "Highest Possible Performance."  That's
> a funny one.  The highest possible performance comes at the highest
> possible price, right?  There are fairly limited resources involved in
> maintaining ZODB.  As far as I know, no one has been motivated enough b=
y
> measured performance problems with ZODB to attempt to fix them.
> That's not saying that ZODB is fast enough for every application;
> rather, it's fast enough for Zope, so Zope Corp. doesn't have much
> incentive to invest time in speeding it up.  I'm not opposed to making
> ZODB faster, but it's not my top priority.

More resources may be put toward ZODB if the importance of it were to=20
increased.

Also wanted to thank everyone for their hard work on open source projects=
 like=20
this !!!!!!


On Monday 02 December 2002 3:57 pm, Jeremy Hylton wrote:
> > ZODB Wishlist:
> > - Distributed
> > - Replication
> > - Data Consistency/Integrity (ACID)
> > - Multiple Storages (physical and or logical)
> > - Scaleable (TB +)
> > - Highest Possible Performance
> > - Easy to use, maintain, and develop for.=3D20
> > - The ability to Query objects.
>
> Nick,
>
> Many of the features you mention are already supported in ZODB3.  Can
> you be specific about what is lacking for "distributed" to pick the
> first example?
>
> Also, I'll have to comment on "Highest Possible Performance."  That's
> a funny one.  The highest possible performance comes at the highest
> possible price, right?  There are fairly limited resources involved in
> maintaing ZODB.  As far as I know, no one has been motivated enough by
> measured performance problems with ZODB to attempt to fix them.
> That's not saying that ZODB is fast enough for every application;
> rather, it's fast enough for Zope, so Zope Corp. doesn't have much
> incentive to invest time in speeding it up.  I'm not opposed to making
> ZODB faster, but it's not my top priority.
>
> Jeremy

--=20
Nick Pavlica
EchoStar Communications
CAS-Engineering
(307)633-5237