Spring dm Server and Eclipse project Virgo

[Johan Pelgrim]

Hi Thierry & Andy,

In your opinion, what has been the adoption level of the Spring dm Server (and maybe OSGI in general) up to this point and how will the submission of the dm Server project to the Eclipse foundation contribute to that?

http://www.eclipse.org/proposals/virgo

Cheers,

Johan Pelgrim
The Netherlands

[Andy Piper]

Johan Pelgrim wrote:Hi Thierry & Andy,

In your opinion, what has been the adoption level of the Spring dm Server (and maybe OSGI in general) up to this point and how will the submission of the dm Server project to the Eclipse foundation contribute to that?

http://www.eclipse.org/proposals/virgo

Cheers,

Johan Pelgrim
The Netherlands

Hi Johan,

That's a good question. Remember that I work for Oracle (ex BEA) so my perspective may be a little biased - and this answer is only my personal opinion

I believe that dm Server has been trying unsuccessfully to compete head-to-head with WebLogic Server (and others). The problem is that it doesn't have the features or
hardening that enterprise customers expect and also the group that may be in favor using of dm Server (the developers) are not the ones making the buying decisions.
I think the fact that dm Server is being donated to Eclipse is probably because it is not gaining sufficient traction.

As to how its donation will change the OSGi landscape, that's really hard to say. Obviously the issue is that there are already other open source servers out there so
customers really need to believe that dm Server delivers superior value in order to choose it over another. That probably means customers already using Spring and OSGi
which must be a very small subset. I think the real value of the Gemini project in general is that it will mean many OSGi components are developed as open source and therefore can be readily adopted by customers and vendors alike. Virgo will no doubt contribute to that process by providing an integrated platform for people to experiment with and
also encourage the development of an ecosystem which surely must be vital to the adoption of OSGi.

andy

[Johan Pelgrim]

Thanks for your answer Andy! Does Thierry have a different opinion?

Cheers,

[Arnaud Cogoluegnes]

From what I heard, dm Server didn't get a large adoption, perhaps because it's been too ambitious on some parts. That's a pity because it allows a smooth transition from the Java standard way to the OSGi way, as it comes with a popular Web container (Tomcat) running under OSGi and its lets you leverage OSGi in your applications.
Hopefully, it will get a broader adoption within the Eclipse foundation, perhaps under a slighty different shape and by narrowing its scope (e.g. removing some proprietary archives like PAR).

[Johan Pelgrim]

Arnaud,

Is maybe what you're saying "OSGi didn't get a large adoption"? I mean, yes, the non-OSGi (deployment) model is flawed, but people've learned to live with it and have their well known work arounds in place and it's "stable". OSGi is probably the better way to do things, but maybe it has come too late for people to eagerly adopt it?

Cheers,

Johan

[Johan Pelgrim]

P.S. Arnaud, sorry for not including you in the addressing my initial post!

[Thierry Templier]

Hi Johann,

Sorry for my late answer! Your question is quite hard to answer . I think using OSGi for applications can scare people since things work a bit differently regarding organizations, classloading and runtime. But this technology allows develop more modular and structured applications and you can benefit of a lightweight service oriented feature and runtime facilities for application administration. Perhaps, one drawback for OSGi use is sometimes the difficulties to identify problems caused by strict classloading...

dm Server tries to address such aspects. I find concepts and implementations of the tool really interesting ... I think the SpringSource label can be a problem to adopters because they probably want tools independent of a company, especially from new concepts like PAR which aren't standardized and part of the JavaEE specification (even if they aim to simply use of OSGi technology)...

A lot of work is done at the moment in order to provide specifications dedicated to enterprise development based on OSGi. I think that creation of Virgo and Gemini projects go in that way...

Hope it answer your question
Thierry

[Johan Pelgrim]

Hi Thierry,

Thanks for your answer. I feel the same way. The promise of OSGi is a good one, and eventually we (should) go the OSGi way, but adoption has been slow in our projects. Maybe because we've learned to live with- or have accepted the current (flawed) classloading mechanism? I certainly agree with you that donating the dm Server project to the Eclipse foundation will have a (very) positive impact on overall OSGi adoption (i.e. no ties to SpringSource). I'll keep an eye on the Virgo and Gemini projects!

Kind regards,

Johan

[Thierry Templier]

You're welcome, Johan!

A link to an interview of Adrian Colyer about dm Server and Virgo: http://jaxenter.com/Adrian-Colyer-Why-dm-Server-Is-Moving-To-Eclipse-10106.html. It surely will interest you...

Thierry

[Johan Pelgrim]

Thanks again Thierry!

Cheers,

Johan

[Arnaud Cogoluegnes]

Johan Pelgrim wrote:Arnaud,

Is maybe what you're saying "OSGi didn't get a large adoption"? I mean, yes, the non-OSGi (deployment) model is flawed, but people've learned to live with it and have their well known work arounds in place and it's "stable". OSGi is probably the better way to do things, but maybe it has come too late for people to eagerly adopt it?

Cheers,

Johan

OSGi has not got yet a large adoption in enterprise applications development, but it is mainstream in other areas like embedded devices. Think also about its adoption in applications like Eclipse. Remember that more and more middleware like application servers (Glassfish, dm Server, Jonas) and ESB (Service Mix) build on OSGi, because it's a very strong modular foundation that comes ready to use.