I understand you mean Service Modeling Language by SML and Choreography Modeling Language by CML. Right?

Thanks!

——
Fermín

PD. Sorry for the previous wrong post (maybe a “Preview” function should be convenient).

I understand you mean Service Modeling Language by SML and Choreography Modeling Language by CML. Right?

Thanks!

——
Fermín

PD. Sorry for the previous wrong post (maybe a “Preview” function should be convenient).

I’ll try to answer a few questions. (Standard disclaimer that I speak for myself here).

> Does he feels the need to narrowly redefine “utility computing” (who knew that all that time “utility computing” was
> just referring to a single hypervisor?)

There’s a difference between “term in concept” vs. “term in practice” — different people have certain connotations associated with an image. I think the Wikipedia entry for cloud computing shows why the new term has emerged, though I agree it remains to be seen if it sticks or if the traditional “utility computing” label is more appropriate.

>Basically it sounds like they don’t want to be seen as competing with Opsware and Bladelogic so they try to redefine > the category.

Of course, that never happens :-) Hello, Workflow -> BPM, EAI -> ESB, Portals -> Blogging Frameworks, Metadata Repositories -> CMDBs, Directories -> Identity Management….

I guess the point is that technology evolves in weird ways that synthesizes prior approaches while it blazes new trails. So, there’s bound to be overlap with that’s come before, but the question is whether the new angle is sustaining the traditional approach or disrupting it? Naturally, cloud computing providers believe there’s a level of disruption at play here, beyond “just another way” to do provisioning & runbook automation. Part of this disruption is economic and has to do with the level of abstraction being offered — holistic services instead of tangled underpinnings. Another part of the disruption is the change in architecture — “everything goes, we’ll discover it for you” vs. one that assumes the Web’s architecture as central. Time will tell how this plays out.

> So where are the specifications for these ECML and EDML formats? I would be very interested in reading them, but
> they don’t appear to be available anywhere. Maybe that would be a good first step towards making them industry
> standards.

I think I’ll echo Bert’s comments that a) other standards aren’t up to the task, and b) one doesn’t want to drive a standard too soon, especially in an evolving space.

Generally building “one modeling language to rule them all” to be a fools errand, especially when many such attempts have largely gone nowhere, and that even in the modestly successful areas, interoperability is minimal. My vote is that “SML/CML” will try to do too much at once, but I’m prepared to be pleasantly surprised.

The current direction of EDML, for example, is as a hypermedia markup language, not necessarily a modeling language. I think that’s a different approach vs. other emerging standards in this space.

In cases where a language is targeted at a similar set of bounded goals, I could see great benefits. I could see a future where we compare notes between 3Tera’s work and Elastra’s work, pick the best union, and together evolve a standard out of that, for example. But it will be some time before that makes sense.

> Maybe Stu will one day explain what a “private cloud” is. His description of his new company seems to confirm my
> impression that they are really focused (for now at least) on “public clouds” and not the Opsware-like “private
> clouds” automation capabilities.

Well, I think to be a “cloud computing provider” you do need to have a firm direction on the “future way of doing things”, which we bet are public clouds, with one’s feet on the ground on “how things are”, which is what private clouds are about.

> Maybe the “private clouds” are just in the business plan (and marketing literature) to be able to show a huge
> potential markets to VCs so they pony up the funds. Or maybe they really plan to go after this too.

We really plan to go after this too, this year, in fact. The work to date has been on an architecture to bridge both public and private clouds, which enables realistic transition strategies.

For example, if you can reduce your basic data centre costs to Amazon EC2 levels, it would be good to exploit these technologies with your own assets. If you require locality of data for security or latency reasons, same deal. As you suggest, one of our major goals is to be able to provision storage and some processing in the private cloud while linking to other services in the public cloud.

We didn’t publish ADL initially for user consumption because to our knowledge no one had ever attempted the type of application level portability that we wanted to accomplish. Rather than manage existing gear, we were looking to package distributed systems into first order entities. Therefore we were concerned that we might have to revamp ADL periodically, causing users to have to rewrite code. ADL was a means to an end rather than a product in itself. After a year plus into production, though, it’s been very stable and we’ve found that users do want to see it and be able to edit it directly so we’ll be making it more visible in the production system.

In this blog entry, I am not saying that ECML/EDML should necessarily be made available as standards. I am just pointing out that they are not, and as such should not be called “standards”. You can’t have it both ways.

I am very interested in learning more about ADL. I find a bunch of links to an ADL specification on your Wiki, but clicking on that link only returns this error: “The page you requested is not available. It has either been removed or it is available only to 3Tera partners”. Which is it? Is there an updated public link or is the spec only available to partners?

I would also be very interested to know what existing specs you considered at the time and what were your key requirements. Assuming you can and care to share more on this.

Having lived through this decusion process, I can tell you that at 3tera, where our ADL (application description language) has been in use for almost two years, we decided not to use other standards because they weren’t up to the task. They specified a great deal of detail but without creating the necesary boudary around an application that enable you to manipulate the end product.

Bert