Interest in joining OpenStack as an incubating project?

[Paul Querna]

Hiya All,

Now that the first release of Cast is out, what are everyone's
thoughts about trying to join the OpenStack community?

The process to do so is outlined here:
<http://wiki.openstack.org/Governance/Approved/NewProjectProcess>

I think it would be a good place to help Cast grow up, and help
increase the size of the community.

If there is consensus, the next step would be to send some mails to
the openstack mailing list and try to get people on their side
interested.

Thoughts?

Thanks,

Paul

[Tomaž Muraus]

Hello,

I like and support the idea.

I think joining OpenStack would definitely help us with exposure and getting more developers.

I also think that Cast nicely aligns with some of the OpenStack goals and it would be a great and valuable addition to it.

[Adrian Otto]

The guys at Gigaspaces have been working on a PaaS solution that they presented today at the OpenStack Summit. It's focus is a bit different than Cast's deployment central focus. I think it would make a lot of sense to get a coordinated effort going with participation from a number of stakeholders in the PaaS space, and see if we can get a shared vision for a collaborative project to bring to OpenStack. Perhaps a good way to start is to invite some key players, such as the CloudFoundry guys, Gigaspaces, our Cast team, and any other suitable participants and have a meeting where we spend a whole day or two together digging into what matters in a PaaS, and arriving at a point where we have a shared vision. Then let's take that shared vision to the Openstack community.

This approach gives us a way to get into deep debate about the best way to set up an open source PaaS, and arrive at some level of alighment prior to having these sorts of discussions in front of the public. It's much more productive to have a sense of unity going into a discussion with members of the public. I have learned this lesson from observing dozens of meetings and conferences over the past few years about cloud computing standards. PaaS is such a wide subject, and there is so much to discuss that you really need to get the basics hammered out in advance.

I am willing to help arrange and host a meeting like this.

Thoughts?

Adrian

[Paul Querna]

On Tue, Apr 26, 2011 at 6:02 PM, Adrian Otto <adria...@rackspace.com> wrote:
> The guys at Gigaspaces have been working on a PaaS solution that they presented today at the OpenStack Summit. It's focus is a bit different than Cast's deployment central focus.

Yes, their approach from what I've seen is completely different.

I kinda feel like they both have their places.

Calling it either a PaaS is a huge generalization trying to get some
'idea' of what each is doing to a user in a single sentence.

> I think it would make a lot of sense to get a coordinated effort going with participation from a number of stakeholders in the PaaS
space, and see if we can get a shared vision for a collaborative
project to bring to OpenStack.

I'd be happy to entertain it, but I kind of believe there is a huge
space we are describing as "PaaS" and there is space in it for many
different projects.

> Perhaps a good way to start is to invite some key players, such as the CloudFoundry guys, Gigaspaces, our Cast team, and any other suitable participants and have a meeting where we spend a whole day or two together digging into what matters in a PaaS, and arriving at a point where we have a shared vision. Then let's take that shared vision to the Openstack community.

There is an assumption that there is even a shared vision there :)

> I am willing to help arrange and host a meeting like this.
>
> Thoughts?

I'm happy to attend such a meeting.... but I'm not sure it has a
direct relevance to the question if Cast should try to be part of
OpenStack. Maybe my view is tainted, but in the Apache Software
Foundation world, it was perfectly okay for there to be two or more
overlapping projects, they are together at the foundation for legal,
community and infrastructure reasons, but the success of an individual
open source project isn't at a detriment of any other.

In my view, the PaaS space is so massive, and Cast is really focused
on delivering APIs into the most common service management tasks and
deployment tasks (install, upgrade, rollback, etc), that it doesn't
really even overlap with the grand vision of what I've seen from
Gigaspaces <http://natishalom.typepad.com/nati_shaloms_blog/2011/04/gigaspaces-openstack-explained.html>

Thanks,

Paul

[Russell]

On the question of joining OpenStack, I think its a great path forward for Cast.

As to the other issue, I'm generally with Paul on this one. The way I see it, my interests in Cast are:

1. To solve the "deployment problem" for our users

2. To do so in a way that anyone can leverage to build higher level (ie, frontend, etc) systems

Obviously neither of those goals is served by needlessly duplicating the efforts of others, so I'm all for trying to hash out exactly what various PaaS solutions are doing and where it makes sense trying to find ways to play nice with the rest of the ecosystem.

On the other hand, from what I've seen Cast thus far is sort of in its own category. On the one hand the vision of Cast is to be able to do some of the things that Cloud Foundry, Joyent (with No.de), Heroku or any number of other framework-specific (whether they're specific to one or several platforms) services are doing. On the other, we're trying to do a number of things that they don't, while they do a number of things that we won't. The overlap isn't, from my perspective, a problem.

But discussion is never a bad thing, and I'm interested to see what others are doing (looking through the slides from the Gigaspaces presentation, I think they do a really good job of breaking down the problem, but its not totally clear to me what they're doing on it), etc.

-Russell

[Adrian Otto]

These are exactly the reasons I think it would be a mistake to register this project with Openstack with "PaaS" as the description. Bringing the larger combination and a sensible explanation of how various tools fit together would be a much more useful contribution to start with. That descriptive explanation that's the product of a collaborative analysis would help tremendously in how this concept is received by the community. I'd love to be able tomake a pitch to the Openstack community that sonds along the lines of:

VMWare, Rackspace, Gigaspaces, Rightscale and (add other relevant thought leaders here) have met to discuss PaaS for Openstaack, and here is what we all agree on...
Here is where current open source projects fit...
Here is where we have problem domains that have not yet been addressed in open software...

etc.

Adrian