[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: *URGENT* -- Call for support
Hi James
I agree with most of things you say.
I simply think that you minimize the impact of JDO on the EJB spec.
JDO is not yet another persistence framework. JDO defines object
persistence, whatever the source is, in 2-tiers or 3-tiers (transactionally
managed) environments, and I think that JDO/EJB containers are really
important.
JDO is important because this is not an ODBMS standard.
It's also important to see what was the original Falko's idea :
* a study project, during spare time, to prove a concept (Java Database)
* something done specifically for a first customer
* a first release of the next database generation
* something else ...
Surely a lot of people can freely contribute, but I think it may be worthy
to coordinate.
Nobody would start an Ozone-JDO interface if someone is already doing it
somewhere.
Best Regards,
Eric Samson
LIBeLIS, Liberty for Large IS
President
+33 6 0302 5341
Eric@libelis.com <mailto:Eric@libelis.com>
-----Message d'origine-----
De : ozone-users-owner@ozone-db.org
[mailto:ozone-users-owner@ozone-db.org]De la part de Trawick, James
Envoyé : vendredi 19 janvier 2001 01:14
À : 'ozone-users@ozone-db.org'
Objet : RE: *URGENT* -- Call for support
Mr Samson,
I am not speaking as a member of the ozone project. I am speaking as an
active member of the IT subculture proclaiming themselves to reside six
months beyond the bleeding edge. I have recently defended against very
similar criticisms among my collegues, and am looking down the barrel of
facing more. Allow me to practice on you, if you will.
> Hi all
>
> Please let me add few things on that thread, as a former technical
Director
> for a well known commercial ODBMS in Southern Europe, and now as the
> president of a brand new company that tries to push Open Source
initiatives
> whenever possible.
>
> * it's quite easy to find people that can contribute to an XML thing, or
to
> an IDE tool, or an JSP something
> * but real database kernel skills is not something so common
I could say the same thing about entrepreneurs and successful startups in
today's market. I hope you get my point. Open source projects rarely claim
to have the best developers in the industry, while they frequently do have
them. The rest of us need to get experience at it, but most of us don't
have the luxury of working at Oracle or Sybase. Here is where we gain
experience and understanding while any holes that pop up are filled in by
the massive independant peer reviewing of the product. Quality loss is
negligible in the long run and next time we run across similar problems, we
draw on our experience.
However, from a management perspective you are right.
> * suppose that Open Source intiatives can take 10 % of a market
> * and today ODBMS is <2% of the whole DBMS market
> * => so the target for an Open Source ODBMS is no more than <0,2 % which
is
> not huge
This depends on your definition of a market. From a corporate standpoint, a
market is defined as the set of potential clients who are willing and able
to pay for a product or service. From this perspective, you're probably
pretty close to the right numbers there when the law of averages are taken
into account.
However, in the world of open source the market is defined as potential
installation base. THIS is what makes open source work. Installations by
students, the technologically curious, and the downright technocrats GREATLY
outnumber the adoption by corporate business models. A data point is a data
point, a mind is a mind, a programmer is a programmer, and the owner of each
installation is going to hack at the thing to see what it can do. That is
something you simply do not get with a closed source model which requires
lawyers and NDAs and mountains of cash to look at the code just to see if it
can be tuned better for their particular situation.
Another dimension to this issue is the fact that if an individual or
corporation is wondering, "What is this object thing and databases?" then
they're usually not willing to spend a lot of money until they're sure that
it represents added value for them and they can adapt to it. So, much of
the time they'll download some sort of free software (be it OSI, trialware,
or freeware) to see what the technology is all about. Given that I don't
know of any trialware ODBMSs (especially those that support persistent DOM)
and such projects are generally too daunting for your average freeware
developer, open source projects such as ozone are prime candidates for such
reviews.
I would guess that the target would be closer to 2% in the short run until
the technology is better proven to the masses, and after that it may yet
steal much of the RDBMS share. However, it will require a new direction and
better marketing leading to industry initiative.
> * analysts says that some technos are ready for Open Source, while others
> are not
> * in the first category they put : Operating Systems, middlewares, ...
> * in the second they put : Databases in general
> * please cf. the latest Forrester study on Open Source technologies
Please see various documents by Eric S Raymond available at
http://www.opensource.org/
> * I'm quite convinced that ODBMS market is having a new birthday owing to
> Java and J2EE
> * EJB market is a huge potential for ODBMS
They are really two different solutions to two different problems. EJBs
pertain to managing resources in a scalable manner and abstracting business
logic from the application. ODBMSs simply provide a persistent object
store.
> * there is NO "xml storage market" (see Excelon troubles these days) !!!!
> * xml is definitely not a storage format, it's only an exchange format
I'm sorry, but I had to chuckle at this one. See Excelon! From the people
who brought you ObjectStore!
But in all seriousness, if you are referring to documents in the classic
sense of the word, then you are right. However, XML persistence may very
well be jumpstarted by industry adoption of XSLT. I know I'm taking a leap
here, but hear me out.
I've been working with XSLT for a few months now and I see how truly
powerful, modular, and extendable it is. For all intents and purposes, it
has taken over all HTML/WML presentation in my JSP. However, that
transformation is driven off of session info, user data, and content from
various sources. Natively, the transformer grabs this data off of a DOM
tree or SAX parser. Alternatively, one can write XSLT extensions to grab
the data from other sources, such as JDBC. The advantage of one over the
other is obvious: writing extensions means that much more code to maintain,
ignoring that the extensions API is difficult to deal with in the first
place. Reading site content data with SAX will be a strangling bottleneck
if the server reaches any volume. Reading user and session data through SAX
is risky at best since that data is unpredictably dynamic, and such updates
would have to be written by serializing the result back to disk on top of
the old file which any number of people are using at the time (the dilemma:
latency or data corruption). Having site content, user data, and session
info resident in memory as a DOM tree will blow the server's memory for any
decent-size site. Thus a compromise, the hybrid known as Persistent DOM.
That is the true power in ozone.
> * ODMG is dead ... (yes it is)
> * ... long live to JDO !
Stinks like a religious issue to me, so I won't touch it.
> What was your original idea/target when you started Ozone ?
> Is there an updated requirement/specification document ?
Stinks like Corporate America to me, so I won't touch it.
> Do you have somewhere a kind of to-do list with needed enhancements,
> priorities, assigned resources, ...
Stinks like a mailing list to me, so I'll leave it at that.
> How many downloads do you have per month (and total) ?
> How many people are regularly contributing to Ozone ?
> How many contributors would be necessary to complete your product, in how
> much time ?
> What are your relationships with Apache, ObjectWeb and jBoss ?
> Are you working with universities ?
(rearranged, yes) I spend about ten minutes trying to fit these into the
context of this posting in an attempt to decypher meaning in this. Are you
saying you want to help? If so then as with any open source project, I'm
sure your resources are more than welcome in ozone. However, if this is the
case then I respectfully suggest a less patronizing delivery. I'm quite
sure that the members of ozone are familiar with resource management,
project networking, and academic support as I'm sure you are familiar with
the troublesome nature of headcounts and schedules as applied to open source
projects without significant corporate backing.
> Falko, I'm quite sure that JDO is the best chance for your product.
> In the next months a huge interest will raise for JDO drivers, and ODBMS
are
> obviously "pure" JDO sources.
Think YAIPS! from falling off a chair after standing on the seat (or Yet
Another Interoperability and Persistence Standard). If you'd like to see
that functionality added to ozone, you are encouraged to do so and free to
say so. If you don't feel the urge to contribute, I'm sure the current
developers of ozone will take your suggestion under consideration and get
around to it if and when it seems worth their time or curiosity.
Welcome to Open Source,
James Christopher Trawick