Re: [opennms-devel] Re: [opennms-discuss] Database Future for OpeNMS?

[David H. <da...@op...>]

On Feb 24, 2005, at 3:37 PM, Miskell, Craig wrote:

> Howdy,
> 	Just saw a message on the discuss list with the above subject,
> and it disturbed me.

Thanks for posting.  This is great feedback, Craig.

>   David H mentioned that the current leaning is
> toward EJB 3.0 as the DB persistence layer.  May I just say that I have
> been here before (admittedly a couple of years ago) and EJB did not
> appear to be a particularly good match, if all you want is DB
> persistence.

EJB 3.0 has come a long way.

>   If you want enterprise level (whatever that means)

Good buzz word... We use it (grin).  There are some valid connotations, 
though.

> distributed objects, scalable to multiple boxes where you don't care
> particularly which box the Bean currently lives on, and all that jazz,
> then sure EJB is a good thing.  If what you want is O/R mapping, there
> are better tools.

Well, we do want O/R mapping and hopefully we won't develop EJB 3.0 
support and not allow for an O/R tool.  We are very big on Hibernate.  
We actually have Hibernate code, not in CVS yet, that already supports 
OpenNMS.  We did it for a custom dev and the OpenNMS project wasn't in 
a stable to safely merge that code in, anywhere.  Perhaps we can make a 
new branch, bring the code in that we have completed and you can help 
us merge it in.  That would be fun!

> This may be too little too late, but may I recommend at least checking
> out Cayenne (http://objectstyle.org/cayenne/).

Looks like good technology and sound very similar to what we did with 
Hibernate and POJOs (EJB 3.0 style, too, btw).

> I apologise if this sounds like a religious rant - I'll freely
> acknowledge it's been a while since I did this stuff as my day-job, but
> when I did, it seemed clear to me that Cayenne was a Good Thing (tm).
> I'll also admit I contributed some code to it at one stage, but that's
> probably been ripped out since, such is the way of things ;-)

We are really leaning forward on EJB 3.0 to for many reasons.  We have 
made agreements with the OpenOSS project to move to J2EE and we already 
had one very large equipment vendor wanting to know if this was on the 
radar.  All good things for the Project.

Another major equipment vendor (the biggest) wants a tool with workflow 
integration. J2EE workflow engines are available and attractive to this 
vendor as well.

What is really keeping this project alive right now is the ability of 
the OpenNMS Group to work on it full time and that requires a revenue 
model built on services.  We won't build J2EE in as a requirement 
unless the community supports it but we see it as a requirement to 
higher end consulting services and partnerships.  The main reason that 
we love working on this project is that we love the work.  "Enterprise 
grade" consulting services not just pay the bills but is what we really 
enjoy.

This was my rant.  My goal it work with the community and build a 
project that we all love (grin).  We are realizing more everyday, 
however, that we really need to get capsd and collectd in much better 
shape before any of the coolness makes much difference.


-David