jsr342-experts@javaee-spec.java.net

[jsr342-experts] Re: Support for the Platform as a Service model

From: Jevgeni Kabanov <jevgeni_at_zeroturnaround.com>
Date: Wed, 15 Jun 2011 21:21:57 +0300

I'm still catching up with the reading and some research, but here are in no particular order some things I'd ideally want to see from this:
1. App servers should abandon the notion that they are only managed by their own consoles. Their should be a generic way to start/stop/deploy/etc. I haven't looked into JSR-77 yet, maybe that's the one. But I'd propose to standardize on the command line instead of the Java API, which is hugely cumbersome in a heterogenous environment the apps commonly live in.
2. Also, please let me pass parameters to the JVM, e.g. managing Glassfish through scripts is impossible. Being able to tune the app server and its JVM centrally is a must for the cloud.
3. REST API layered over JMX API for deployment, configuration management and app management. Can't stress that enough.
4. A jvm-instance-per-app model (better called process per app model) for app servers as an alternative to the current classloader per app model and way to manage those apps. At least as an option to be standardized on down the line.
5. Session API/SPI (probably through the same JMX/REST combination) that allows to migrate data without relying on the app server and also keep session in the super-effective datagrids/caches.
6. Oh, and an API for provisioning app servers, though that's probably too much to expect :)

Generally I think the main thing PaaS support requires is to expose a bunch of scripting APIs and make the app server consoles to be consumers of the APIs the underlying server exposes. This will create a great ecosystem to complement the existing tooling instead of requiring app servers to catch up one by one.
--
Jevgeni Kabanov
Founder & CTO of ZeroTurnaround
http://twitter.com/ekabanov
On Wednesday, June 15, 2011 at 18:47 , Reza Rahman wrote:
> Jevgeni,
> 
>  I agree. I think this is really the practical way most multi-tenancy would be implemented.
> 
>  Cheers,
>  Reza
> 
> 
>  On 6/15/2011 8:40 AM, Jevgeni Kabanov wrote: 
> > Personally, I don't think that modularity will give any advantages to PaaS or multi-tenancy. For PaaS using process-per-application model would make the most sense, as that would provide proper isolation and for multi-tenancy either separate JVM instance can be used to ensure isolation or a monolithic approach to leave isolation requirements to the app developer.
> > 
> >  -- 
> > Jevgeni Kabanov
> > Founder & CTO of ZeroTurnaround
> > http://twitter.com/ekabanov
> > 
> > 
> > On Wednesday, June 15, 2011 at 15:34 , Antonio Goncalves wrote:
> > 
> > > It's scary to thing multi-tenancy without having modularization and wait until SE 8 (Jigsaw). I really wonder how we will get trough this PasS/SaaS adventure without modularity and just guessing what will happen (even if Jigsaw is pretty much defined) 
> > > 
> > > Antonio
> > > 
> > > On Tue, Jun 14, 2011 at 20:22, Adam Bien <abien_at_adam-bien.com (mailto:abien_at_adam-bien.com)> wrote:
> > > > 
> > > > On 02.06.2011, at 21:42, Jeff Genender wrote: 
> > > > > Im in agreement with a possible cloud profile, but I don't think putting off a PaaS item until JavaEE 8 is a good idea. 
> > > > 
> > > > +1. We should at least prepare things and then wait for Jigsaw :-) 
> > > > 
> > > > > The rate which we get things out, there will be a next big thing. Lots of folks are clamoring for PaaA standards and I think we are in a good point to have it a part of this spec. 
> > > > > 
> > > > > Jeff 
> > > > > 
> > > > > On Jun 2, 2011, at 1:36 PM, Reza Rahman wrote: 
> > > > > >  Antonio,
> > > > > > 
> > > > > >  Glad I'm not the only one seeing things from this perspective :-). Quite honestly, I would not be at all opposed to delaying the entire multi-tenancy idea to Java EE 8 or later and focus on the more "bread-and-butter" things that need to be done :-).
> > > > > > 
> > > > > >  Cheers,
> > > > > >  Reza
> > > > > > 
> > > > > > 
> > > > > >  On 6/2/2011 1:47 PM, Antonio Goncalves wrote: 
> > > > > > > I'm following reza's comments on "not everybody needs cloud" and I think we should have a Cloud profile instead of having it on the EE platform ? Profiles can be subsets of EE as well as super sets. So why not having a Cloud profile that includes most of EE and adds extra cloud features ? This way we could also leave EE as it is (which means no Cloud, Cluster, multi-tenant...). What do you think ?
> > > > > > > 
> > > > > > >  Correct me if I'm wrong, but latelly I've been earing a lot about multi-tenant. It looks to me that the world is moving the other way now. Application servers, databases... tend to be lighter and lighter in terms of resources. So why bother having  multi-tenancy ? Just instanciate several servers or DBs and each instance only hosts one application. If I'm not wrong (Nicolas Leroux can correct me) a Play! instance takes less than 100Kb or RAM. With profiles, OSGi and modularity arriving, the platform is  shrinking and the app servers will shrink too, so why not goaling for several app servers with one application instead of one app server with several applications (same for DBs) ? That's why I think all these PAAS concepts could go into a profile, not EE. Multi-tenancy is useful for some people, but not everybody.
> > > > > > > 
> > > > > > >  My 2 cents
> > > > > > > 
> > > > > > >  Antonio
> > > > > > > 
> > > > > > > 
> > > > > > > 
> > > > > > > On Sat, May 28, 2011 at 1:02 AM, Reza Rahman <reza_rahman_at_lycos.com (mailto:reza_rahman_at_lycos.com)> wrote:
> > > > > > > > Bill,
> > > > > > > > 
> > > > > > > >  OK, all of what you said makes perfect sense - thanks for the prompt response.
> > > > > > > > 
> > > > > > > >  We'll put some more thoughts on some of this internally and will try to share helpful details.
> > > > > > > > 
> > > > > > > >  Cheers,
> > > > > > > >  Reza 
> > > > > > > > 
> > > > > > > > 
> > > > > > > > 
> > > > > > > >  On 5/27/2011 6:20 PM, Bill Shannon wrote:
> > > > > > > > >  Hi Reza, welcome back! Linda is busy with JPA today so let me try to
> > > > > > > > >  answer your questions...
> > > > > > > > > 
> > > > > > > > >  Reza Rahman wrote on 05/26/11 02:36 PM:
> > > > > > > > > >  Linda,
> > > > > > > > > > 
> > > > > > > > > >  Overall, this is a good start. At the same time though let me state up-front two
> > > > > > > > > >  concerns:
> > > > > > > > > > 
> > > > > > > > > >  Firstly, I hope all of this is implemented in away that does not effect
> > > > > > > > > >  developers of simple applications that do not require cloud support and never
> > > > > > > > > >  will.
> > > > > > > > > 
> > > > > > > > >  Yes, that's our goal.
> > > > > > > > > 
> > > > > > > > > > Secondly, I hope the cloud support does not take up so much bandwidth for
> > > > > > > > > >  the Java EE 7 JSRs that more mundane but equally/more important things get put
> > > > > > > > > >  in the back-burner.
> > > > > > > > > 
> > > > > > > > >  Right now I'm more concerned about the reverse. We have a pretty good handle
> > > > > > > > >  on what's required for all those more "mundane" things, whereas "cloud support"
> > > > > > > > >  is still pretty amorphous. There's a tendency to work on the problems we
> > > > > > > > >  understand instead of the hard problems we don't yet understand. Clearly it's
> > > > > > > > >  the latter where we're looking for the most help from the expert group.
> > > > > > > > > 
> > > > > > > > > >  Specific comments on the document:
> > > > > > > > > >  * I was left wondering about the specifics of how a tenant ID get's established
> > > > > > > > > >  in the first place. While it might not be possible to spec that out completely,
> > > > > > > > > >  it might be a good idea to have some guidelines so that vendors don't diverge
> > > > > > > > > >  far beyond effective future collaboration.
> > > > > > > > > 
> > > > > > > > >  Right now my feeling is that there's a number of different ways this could
> > > > > > > > >  be done, and it won't matter to the application which one is chosen, so there's
> > > > > > > > >  no need to overly constrain the solution. If you have a specific scenario
> > > > > > > > >  that you're worried about, let us know.
> > > > > > > > > 
> > > > > > > > > >  * I think it is best not to make cloud support the default platform behavior but
> > > > > > > > > >  rather something that is enabled specifically. If this really becomes cumbersome
> > > > > > > > > >  in the future because a majority of applications are on the cloud, we can always
> > > > > > > > > >  make it the default later. Going the other way round if cloud computing turns
> > > > > > > > > >  out to be just another over-hyped, vendor-driven bust with limited practical
> > > > > > > > > >  applicability is going to be difficult I think.
> > > > > > > > > 
> > > > > > > > >  Our current thinking is that an application is going to have to explicitly
> > > > > > > > >  say "I'm designed for the cloud environment". When we understand everything
> > > > > > > > >  that that implies, we might change our mind.
> > > > > > > > > 
> > > > > > > > > >  * I prefer readable Java identifiers to abstruse UUIDs :-).
> > > > > > > > > 
> > > > > > > > >  We want a tenant ID to be usable as a database primary key, a file name, etc.,
> > > > > > > > >  so I think we only need to constrain the ID sufficiently to make it usable
> > > > > > > > >  in this way.
> > > > > > > > > 
> > > > > > > > > >  * I definitely think cloud support should be optional or at least not added to
> > > > > > > > > >  the Web Profile.
> > > > > > > > > 
> > > > > > > > >  Yes, we expect most of the cloud support to be optional for the entire
> > > > > > > > >  platform. A Java EE 6 product that provides no cloud support should be
> > > > > > > > >  able to be updated to support all the other things in Java EE 7 without
> > > > > > > > >  also having to explicitly support the cloud environment. It may need to
> > > > > > > > >  understand things about the cloud environment so that it can safely ignore
> > > > > > > > >  them, or provide nop or trivial implementations of them, but it shouldn't
> > > > > > > > >  be required to actually run in a cloud environment.
> > > > > > > > > 
> > > > > > > > > >  * It's difficult to make a call on ignoring the cloud settings without looking
> > > > > > > > > >  at the overall cloud solution in detail. For now, I would say implementations
> > > > > > > > > >  that do not support the cloud should simply ignore the cloud settings. This
> > > > > > > > > >  would also make development on local machines that need not support the cloud
> > > > > > > > > >  easier while the application can maybe later deployed to a cloud enabled server
> > > > > > > > > >  for testing, production, etc.
> > > > > > > > > 
> > > > > > > > >  Right, we need to get further into the details before we can decide this.
> > > > > > > > > 
> > > > > > > > > >  * Multi-tenancy comes in a very wide array of facets -- the same application
> > > > > > > > > >  deployed to different machines with tenant-specific configuration talking to a
> > > > > > > > > >  tenant-specific database, Multiple tenants using the same application but going
> > > > > > > > > >  to tenant-specific databases, multiple tenants using the same shared database,
> > > > > > > > > >  and so on. It would be important to get those details hashed out centrally and
> > > > > > > > > >  propagate it to the different JSRs as opposed to different JSRs coming up with
> > > > > > > > > >  their own solutions. In this case, if we don't do that chaos might ensue :-).
> > > > > > > > > 
> > > > > > > > >  Yes, that's the kind of thing we'll need to discuss further.
> > > > > > > > > 
> > > > > > > > > 
> > > > > > > > >  -----
> > > > > > > > >  No virus found in this message.
> > > > > > > > >  Checked by AVG - www.avg.com (http://www.avg.com/)
> > > > > > > > >  Version: 10.0.1375 / Virus Database: 1509/3663 - Release Date: 05/27/11
> > > > > > > > > 
> > > > > > > > > 
> > > > > > > > 
> > > > > > > 
> > > > > > > 
> > > > > > > 
> > > > > > >  -- 
> > > > > > >  Antonio Goncalves 
> > > > > > >  Software architect and Java Champion
> > > > > > > 
> > > > > > > Web site (http://www.antoniogoncalves.org/) | Twitter (http://twitter.com/agoncal) | Blog (http://feeds.feedburner.com/AntonioGoncalves) | LinkedIn (http://www.linkedin.com/in/agoncal) | Paris JUG (http://www.parisjug.org/)
> > > > > > > No virus found in this message.
> > > > > > >  Checked by AVG - www.avg.com (http://www.avg.com/)
> > > > > > >  Version: 10.0.1375 / Virus Database: 1511/3675 - Release Date: 06/02/11 
> > > > > 
> > > > 
> > > > Independent Consultant, Speaker, Java Champion 
> > > > 
> > > > Weblog: blog.adam-bien.com (http://blog.adam-bien.com)
> > > > press: press.adam-bien.com (http://press.adam-bien.com)
> > > > eMail: abien_at_adam-bien.com (mailto:abien_at_adam-bien.com)
> > > > twitter: twitter..com/AdamBien (http://twitter.com/AdamBien)
> > > > Mobile: 0049(0)170 280 3144
> > > > 
> > > > Author of: 
> > > > "Real World Java EE Night Hacks", "Real World Java EE Patterns--Rethinking Best Practices"
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > 
> > > 
> > > 
> > >  -- 
> > >  Antonio Goncalves 
> > >  Software architect and Java Champion
> > > 
> > > Web site (http://www.antoniogoncalves.org) | Twitter (http://twitter.com/agoncal) | Blog (http://feeds.feedburner.com/AntonioGoncalves) | LinkedIn (http://www.linkedin.com/in/agoncal) | Paris JUG (http://www.parisjug.org)
> > 
> > No virus found in this message.
> >  Checked by AVG - www.avg.com (http://www.avg.com)
> >  Version: 10.0.1382 / Virus Database: 1513/3705 - Release Date: 06/15/11