[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Request for experiences/advice
David,
Peter is correct, OpenNMS.org is definitely investigating the use of Ozone
instead of a standard RDBMS such as PostgreSQL or even Oracle. Were busy
writing our management platform in java to increase the portability and to
speed up development. It also helps that network object lend themselves to
the concept of "objects" very easily. Part of the problem we struggled with
was trying to represent the network elements in a standard RDBMS and make it
useable.
Coming up with the right set of tables for usability and extensibility
turned out to be a real nightmare. The goals seemed to be mutually
exclusive, we kept searching for alternatives. Not being database "Wonks"
did help either. Shane O' kept coming back to the concept of an object
database (for those poor souls that have worked with HP OpenView that's what
they call their database, however it's not even close to the Ozone
definition IMHO). That's when we stumbled across Ozone and thought this
might really save us in terms of time to market and ease of use.
Here are the advantages I see with Ozone:
1) You don't years of experience using databases to start using Ozone. Any
competent Java programmer can make use of ozone just by learning how to
interface by writing client interfaces and server objects. Since Ozone
stores java objects you don't have to learn another language like SQL to
make use of it.
2) By making sure to use serialization information in your database classes,
the 'serialVersionUID' classes can change over time in the database.
3) It's in java so it in essence "platform neutral." Depending on the
features used in the Ozone Database it could also be used for embedded
systems.
4) The objects live in the server, not in the client. I don't understand it
all in terms of how it affect performance, but it should help consistency
since the server is the exclusive owner of an object.
Your wondering about the support level of Ozone. From what I can tell from
the very limited time we have been evaluating the technology it appears to
be well supported from the development staff. The response times have be
very good w/respect to my queries.
The current downside is that there is no exported (anonymous) CVS access to
the source tree. Nor is the a daily updated tarball available from the
website to the best of my knowledge. So keeping up to date with any
significant changes will be difficult. I've expressed my reservations about
searching/debugging code that may have already been fixed. Considering that
I supposed to focus my efforts on OpenNMS I do not want to waste
brain/computing cycles on a resolved problem in Ozone. However, Shane & I
have agreed that we see Ozone as critical to OpenNMS' success (look for an
announcement from Shane soon on the OpenNMS list) our team will be spending
significant time in the next few weeks to evaluate and implement Ozone in
our network management product.
In terms of documentation for Ozone, we'll from my searching it's scarce at
best. This is the downside of a new project that hasn't yet reached their
1.0 release. OpenNMS is no better in terms of documentation. Everyone is
focused on getting the code to work, adding the required features, and
trying to obtain 1.0 before scope creep pushes it out of reach. For now I
would suggest that you accept that the best documentation you will get is
from questions on the mailing list.
I don't know the size of your application or what your team size may be, but
I would evaluate Ozone on the long term merits of what it can provide. If it
is of value then I would suggest that you get managements approval and see
if you can staff a warm body or two to help out the ozone project team.
Testing, coding, and documenting will not only make the resulting project
more useable for you, but for the community in general. If Ozone turns out
to solve our problem at OpenNMS I wouldn't be surprised several of our
developers end up spending a significant percentage of time helping the
ozone team in addition to working on OpenNMS.
I'll try to keep anyone who is interested up to date on our progress at
OpenNMS w/respect to Ozone. If your interested you might want to monitor the
database mailing list for OpenNMS. 'Nuff Said :)
Best of luck.
Weave
mailto:weave@opennms.org
-----Original Message-----
From: Peter Newell [mailto:ppnewell@bellsouth.net]
Sent: Friday, May 05, 2000 11:59 PM
To: ozone-users@ozone-db.org
Subject: Re: Request for experiences/advice
David,
I know the opennms.org is looking at this, they are a network management
group and a object oriented database fits perfectly. Shane O'donnell is
doing the review his email is shane@opennms.org he may be able to tell you
where they are with this.
At IBM, we created an OO based RDMS and if done right work fairly well, but
not as good as a true ODB.
I would think if you're using a middle layer this would help. Not sure how
well ozone would work with a middle layer, might need to explore that some.
Let me know how you make out.
Peter Newell
----- Original Message -----
From: "David K. Kim" <david.kim@painconsult.com>
To: "'Ozone-Users'" <ozone-users@ozone-db.org>
Sent: Friday, May 05, 2000 10:33 AM
Subject: Request for experiences/advice
> Hello All,
>
> I am evaluating Ozone for use in a commercial
> product (a medical expert system for cancer pain
> management).
>
> Ideally, much of the logic would exist as Java
> middleware, and I am trying to avoid using an RDBMS
> on the back end (e.g. with a poor-man's Object-
> Relational mapper).
> An ODB would seem to make better use of
> development time in that persistence would occur
> more naturally.
>
> My concerns are
> -fewer programmers in the U.S. with ODB skills
> (a concern for longevity and maintenance)
> -commercial systems are costly
> -I can't tell how mature and well-supported Ozone is
>
> I would love to use Ozone since I think it's a
> great project. However, I need to understand better
> the level of support before betting someone else's
> (my company's) resources on this.
>
> Would anyone be so kind as to share their experience
> using Ozone in this way?
> e.g. -Robustness/bugginess
> -Documentation (yes, I know, I know...)
> -Overall ability to "transparently"
> integrate with development work
> -Performance (esp. with a LocalDatabase)
>
> I would be happy to summarize if desired...did not see
> this issue addressed in a FAQ or Listserv.
>
> Thanks,
> Dave Kim
>
> ------------------------------------------------------------------
> David K. Kim, MD david.kim@cynergygroup.com
> Senior Vice President, Technology voice (206) 855-8026
> Cynergy Group fax (206) 855-0851
> "Bringing evidence-based clinical expertise to the point of care."
>
>
>
>