As Novell's Service Line Principal of Open Source and Linux, Carl Drisko has a great vantage point on the Grid / Linux relationship in industries such as financial services - where high performance commodity computing has become mainstream. This month, the Globus Consortium Journal got Drisko's latest thoughts on the specific reasons why Linux is the most widely used OS for Grids, the relationship between the Grid and Linux communities, and Novell's efforts in Linux virtualization.

GCJ: Why is Linux currently the preferred platform for Grid computing?

Drisko: From a performance standpoint, when you're working with Linux, you get very close to the metal. So there isn't a lot of systems overhead, and you can take full advantage of the hardware that you've got, as well as support customization. For example, there are Grid users out there that are accessing a bunch of sensor data. They need to be able to very easily write their own device driver to go and grab some data off of potentially unique devices.

Grid users often aren't just buying solutions off the shelf, they're building them themselves. Linux allows much more flexibility to do what you want, to create custom installations. If something isn't there, you can more easily add it in and make it work better on your own - which is something that's attractive to a lot of people, particularly researchers. Despite all the talk about Grids from some of the major hardware players, there's not a heck of a lot of good software out there, so you still have to do a lot of work yourself to fill the gaps.

Another obvious factor for Linux's popularity in Grid environments is cost. Anytime you're deploying 100 or 1,000 or 10,000 boxes, licensing costs become a big consideration. Cost may not be a big consideration for the financial services world - but for any research institution, where Grid traction has been strongest to date, cost is a huge consideration. Beyond the O/S, it is just as big a factor, if not bigger, for other "up-the-stack" software. And in many cases, the proprietary software (and even some open source) licensing models have not adjusted to Grids.

GCJ: What about SUSE Linux? What are you guys bringing to the table for Grid users?

Drisko: SUSE offers several excellent management tools. If you're deploying lots of servers on a Grid, you run into some specific challenges. For example, you can't have any sort of interference between the various applications that might be running. SUSE has an enhancement called AppArmor, which is a security container for applications.

The Grid community is also very actively working with virtualization today, and in September of last year, SUSE became the first Linux distro to officially pre-configure our kernel with virtualization capabilities. In terms of enterprise support, we've been working very closely with the XenSource teams and with the various hardware vendors - Intel and AMD - to also support the appropriate virtualization technologies that they're bringing out in hardware. So we were the first ones out with built-in virtualization, and we have a lot more experience building out virtualization test systems and working with the chip makers than any other Linux distro.

GCJ: While Linux is the commonly used OS for Grids - the Linux and Grid communities are somewhat disjunct. Various industry pundits have wondered aloud what sorts of opportunities might be afforded if the two sides (Linux and Grid communities) were to more formally come together. What do you think about that?

Drisko: That's an interesting question. I'd be curious to know from the Grid community whether there's anything that the Linux group - kernel.org - hasn't adopted, that they'd really like to see pushed. To my knowledge, there haven't been a lot of demands on the Linux community from the Grid community, so it would be nice to open up the dialogue. But the Grid community has certainly embraced Linux... Linux has been the reference OS for everything that's been done with the Globus Toolkit in the past, for example.

I think from the other side, from the Linux side, really at a high level, those folks are looking towards applications. And the million dollar question is, "what can take advantage of a Grid framework?" I think once that question has been answered, we're going to see the Linux community dive into the Grid space a lot more.

One possibility that's intriguing is a Linux / Grid stack that includes a number of different components that work well together, that make it easy to deploy Grids -- so organizations can pre-configure a lot of things that otherwise require too much assembly and hassle. So I think there's plenty of synergy there -- where many of the different Grid pieces and libraries and other things could be compiled and then released as a Grid stack that can be more easily maintained. Because a lot of the issues that people have with Grids involve scenarios where they change one library over here, and three other software components break down over there. The Linux community has great technologies for making sure that you can retest configurations, and make sure things don't break down when you put in a new patch or change. I think we can accelerate the deployment for the community in the Grid space fairly nicely.

I'd also point out that one area where you do see very good interaction between the Grid and Linux communities is at the standards level. Both of us are very interested in seeing good standards develop across management functions as an example. So I think there are people from both communities participating in those standards bodies helping to establish what features are really needed and required.

GCJ: Would it ever make sense to dumb down the Globus Toolkit for more enterprise-ready installation, just to get more people in the commercial space using it?

Drisko: You certainly don't have to dumb it down, in the sense that you don't have to take out any features. But what would be most helpful would be to provide stylized deployments, where you determine fairly common configurations and people no longer have to make 1,000 decisions to do the deployment. For enterprises, you need to distill it down to the half-dozen essential questions, then set everything else up for them based on that minor set of questions.

A plug for SUSE -- one of the things that we do allow for is the ability for our users to put those types of controls into YaST ("yet another setup tool"), one of our setup tools. Putting a YaST front-end to the Globus Toolkit would be a relatively easy thing to enable and would let the user pre-configure a few choices, then automatically be compatible with the Linux components. It doesn't take that long to do. We've had YaST deployments for things like DB2 as well as all the open source tools. It makes the set-up much easier.

GCJ: Are you seeing any momentum for the "P" (Perl / Python / PHP) languages, as they play into the Grid space?

Drisko: The speed and the ease of development is certainly a major plus in favor of things like the LAMP stack and using AJAX and building composite applications as a whole.

I think it plays very nicely into the Grid space. With Grids, you want to be able to plug into a set of useful services and then be able to have anybody use them. With Grid, here's a built-in infrastructure that I can tie into. With LAMP application infrastructure - boom, I can develop my code in a hurry, scaled out in a way that's much easier to manage. I think in the short-term, you'll primarily see action in the convergence of Grid and LAMP from the companies who are doing Internet-based applications. Obviously some of these stacks are very much geared towards those Internet composite applications.

But I would say, more for heavy-duty database-related apps that call for more transactional support, the J2EE framework is going to be around for a while.

close window