Video Slashdot Talks with David Nalley About Apache CloudStack (Video) 13
Tim: David, talk a little bit about what Apache CloudStack is. Obviously, there are tons of software offerings, let’s say, with the word Cloud in it. Apache CloudStack starts with Apache. I want you to talk about the significance of that.
David Nalley: So Apache CloudStack is a project that is at the Apache Software Foundation. It’s a top-level project there. And of course, that means the Apache Software Foundation provides a home from a legal perspective. It also provides a little bit of the culture, the Apache way that folks have solemnized over a period of time. And it kind of sets norms for our community both from what we produce in a legal sense as far as releases as well as how we operate the transparency, the meritocracy—all of that comes with being an Apache project.
Tim: Now there are a couple of different levels of contributors, people who actually have quite a bit of interest in open-source project, but talk about the organization of CloudStack a little bit. How many people are there involved at each of the big levels?
David Nalley: So, there are essentially three levels. The first level is developers, who are contributing code or documentation or some other form of contribution to the project. When we talk about those 169 people, I think, who contributed to the code base in the past 12 months. There are other people who have contributed in other means. The level above that though is to be a committer which essentially means you’ve the right to commit directly into the repo. You can push there as opposed to having to submit a patch for review. After that of course comes the body of PMC, which is the next level up. It lets committers, after they’ve earned some trust and earned some merit in the projects. So you can’t get it because of your employer, you can’t get in because you are special and awesome. You have to prove yourself to the community over time. And then the level above that is the project management committee. And that particular body has binding ______for decisions and for things like releases, so when we decide to release the software, the people who have binding input to that decision are the PMC numbers. And there are – those two are responsible ultimately for the health of the project, ensuring that it’s continuing to move forward and setting guidance etc.
Tim: Explain a little bit what is your role within the project.
David Nalley: So, I’m a project management committee member and I’ve been – I originally came from Cloud.com before the acquisition, before the move to the ASF, I was involved with the project then. When we made the move to the ASF, I ended up coming on the project management committee.
Tim: Would you clarify for people who don’t know the project very much that acquisition with Citrix.
David Nalley: Yeah. So, Cloud.com started. They were writing software. It was open source, but it was under the GPLv3 license at the time. And that company was acquired by Citrix. Citrix then changed the licensing to the Apache Software license and then shortly, thereafter proposed that could be entered into the Apache incubator.
Tim: And then it became a top-level project which is a fairly small number of the Apache Software Project _____?
David Nalley: So, there are – I think right now, there are 151 top-level projects at the Apache Software Foundation. We are one of those incubations, it took us around a year. So we were in incubation; essentially that doesn’t reflect on the quality or maturity of the code as much as it does onhave you gone through all of the software and made sure it conforms to the IP policies that the foundation has. Have you demonstrated that you are able to manage the project in a way that’s acceptable to the foundation? Have you grown the community so that there is true diversity in the community and it’s not a single entity that controls it?
Tim: Now, let’s talk about the software itself. CloudStack lets people create a certain kind of cloud services for themselves. It’s a little bit different. So, maybe distinguish it somewhat let’s see from OpenStack.
David Nalley: Sure. So the difference between CloudStack and OpenStack, is really kind of design principles. So CloudStack and OpenStack were first written in different languages. OpenStack was written in Python, CloudStack was written in Java. OpenStack is also a collection of services that are distinct, discrete modules, and so you can buy collections of those to build a service. The scope of OpenStack is also much greater. So they have services like object storage. And the scope of CloudStack is much narrower and rather than having collections of modules, we have a little more monolithic model that’s made a little more flexible with the plug-in system but it’s still rather monolithic in comparison to OpenStack.
Tim: Being monolithic does that mean it’s a simple process to install it on your hardware?
David Nalley: So setting up a cloud is never an easy process, right? You are orchestrating compute resources storage and networking—that’s never going to be as easy as installing say OpenOffice. At the same time I think there are fewer moving pieces in a CloudStack install and I think that makes it a little simpler, but it depends upon what your end goal is, I think. Those are two different approaches, neither one is necessarily bad, just different design philosophies.
Tim: One of the things you just mentioned is being available as a sort of a built-in or default thing in OpenStack that is handled separately in CloudStack is object storage, as an example, how would someone who wants to implement object storage do it but they’re using CloudStack instead?
David Nalley: So CloudStack doesn’t provide object storage at all if you wanted to use object storage with CloudStack, so CloudStack can use make use and consume object storage and it supports the S3 API... we support the OpenStack Swift API. So if you set up Swift, CloudStack can consume that object storage. If you set up something like Ceph and expose their S3 API support, CloudStack can easily consume that, and so where consumers of object storage we don’t provide it all, that interaction Swift is a perfectly valid object storage model, and we happily consume that as well as couple of others out there, so
Tim: So you focus on the computation
David Nalley: We do. Our focus is very tight on compute, and so that really is just CPU and RAM. We can’t get away from connectivity, so you got to have some networking and then we also have a need for storage to consume storage, not to provide it per se. And so we focus on interacting with network and compute as well, so that’s hypervisor _____ bare metal, technically, are containers and that gives us – that much smaller scope I think makes things a little bit easier on it.
Tim: Let me ask you about hypervisor support, talk a little bit about what hypervisors are available as options?
David Nalley: And so for the current release of CloudStack we support the VMware we have the center, we support XenServer, KVM, we will support LXC and Hyper-V as well. Neither was one I was forgetting this because I am at a Linux stall, I can’t mention a Microsoft product. In the upcoming release some folks have worked on adding in support for O VM-3. O VM-3 which is Oracle VM manager, and that should be in the next release of CloudStack and there are some work going on right now around Docker in trying to figure out how to use software in the right way, using it just as a container is pretty easy, but we have that kind of functionality with LXC. Figuring out how the model benefits from Docker, in an infrastructure as a service platform API is something we’re trying to get right, if we don’t want to get that wrong then have to support it for years down the road. But that’s kind of where we are from a hypervisors standpoint. The non-hypervisor that we support is also bare metal, and so if your hardware has IPMI support or your Cisco UCS we can orchestrate that as if they were hypervisor.
Tim: Just one more thing, can you talk about some people we’re using CloudStack right now. There are some pretty big names that have used CloudStack installs.
David Nalley: There are. Tell me where you want to start, so obviously there is telecom, be that British Telecom, Alcatel-Lucent, there’s Orange has – Orange in France and Orange in Switzerland have installed, Korea Telecom has a massive ______ViaTel has a large deployment. Then you move into more of the private cloud and so we’ve seen companies like Disney deploy CloudStack in a couple of places. We’ve seen some of the larger gambling houses and so like Paddy Power. Paddy Power is a gaming provider based out of Ireland and if you’re placing bets or gambling on their site you are doing that all through CloudStack.
Re: (Score:2)
I was at that conference. IBM/Softlayer is built entirely from a fork of cloudstack they brand inhouse as IMS.
Yeah, if this is the IMS you're talking about [wikipedia.org] it has nothing to do with cloudstack. Unless you're suggesting IBM forked cloudstack back in 1966.
Re: (Score:1)
Softlayer was only recently acquired by IBM and has been using CloudStack since before the acquisition. Here's an article mentioning Softlayer's IMS (Infrastructure Management System) and CloudStack:
SoftLayer Brings Hourly Billing To Bare Metal Cloud [enterprisetech.com]
Open source: (Score:2)
Open source: so you can host it yourself, make changes you need, and see how it works.
Cloudstack: so you can ignore all that and let IBM take care of you like a little baby bird.
Re: (Score:2)
CloudStack is open source, owned by the ASF, and you can totally host it yourself and make any changes you need.
Citrix has their Citrix CloudPlatform product based on the Apache CloudStack (ACS) source code, IBM might have their own product based on ACS (or SoftLayer), but at the end of the day, neither IBM's product or Citrix's product is ACS as neither company owns or controls the Apache Foundation.
Citrix keeps XenServer and CloudStack on a leash (Score:1)
While CloudStack is under an opensource license, Citrix is gaming the system to appear open while keeping a leach on what functionality goes into the mainstream form of this "open" project.
David Nalley brings up that the benefits of CloudStack being an Apache project is that it provides transparency and makes sure that no one company dominates the project. Both of these claims are misleading.
In reality, there is not the type of transparency you would like to have for something claiming to not allow any one
Re: (Score:1)
It's quite easy to get Apache CloudStack up and running. There are a few Docker containers for the ACS management server as well as Ansible Playbooks and Chef Cookbooks to deploy it for you, but the manual installation process can realistically be completed in less than an hour (provided a local mirror for things like System VM templates and the RPM/DEB packages).