We have had an exclusive interview with Rich Wolski, the founder of Eucalyptus System -the company behind Open Eucalyptus which was ultimately formed to continue a research work which was begun at the University. Let's discover the project, the story behind and which features have made an open-source cloud project of it.

How was born Open Eucalyptus?
Eucalyptus started as an open-source research project in my research group in the Computer Science Department at the University of California, Santa Barbara.  My graduate students, postdocs and I had been studying high-performance distributed systems as a research topic, some for many years by building systems and then observing their behavior under user load.  Our research about cloud computing started when some users began asking us about how they could use a new form of distributed system.

We needed a cloud system we could work with for which we could have access to the source code, and about which we could publish papers without intellectual property constraints.  These two requirements essentially mean that we needed to work with an open-source cloud platform.  (...)
Thus we embarked on a "clean slate" development effort with the goal of designing and then building a cloud platform for use as a "real world" research vehicle (...).  We studied the cloud model and designed a scalable, hierarchical system that we believed we could implement using open-source web-service technologies and open-source hypervisors for machine virtualization.

Immediately, though, we were faced with the problem of cloud API design.  We designed our own API at first but then noticed that much of the academic and commercial discussion cloud computing seemed to be centering on the API and not on the architectural or system response characteristics.  Indeed, we encountered some resistance to our research approach based on the observation that the API we were thinking of did not resemble the APIs of some of the legacy systems that were being repurposed as cloud systems. Since we believed that cloud computing was different at the architectural and not just the API level, we decided that we would side-step the API question by implementing a API universally accepted as a cloud API: the Amazon AWS API.

These events then, formed the genesis of the project.  We conducted the work as part of an NSF funded research effort called VGrADS with the goal of supporting a large, distributed weather forecasting application that had been adapted to run in AWS and we were able to demonstrate its initial viability at a large, high-performance computing conference called "SC."  The rest of the development and open-source community engagement effort stems from this project.

Do you know how many companies use Open Eucalyptus?

We are not exactly sure since many of them either download it and run it on their own or get it through the various Linux distribution channels and never contact us.   We hear from thousands of users in the open-source community, many of whom appear to have commercial email addresses, but we do not track or exploit directly our open-source contacts for commercial purposes.  Many companies (literally hundreds) do contact us to discuss our commercial offerings but we maintain the open-source system for free and unfettered usage.

Among these company is there any large corporation use Open Eucalyptus?
We can't really talk about our large-scale commercial users publically [...]. The largest production deployment we know of, however, supports thousands if not hundreds of thousands of web users.

From your point of view, what brings the open-source distribution to Open Eucalyptus? How do you contribute to open-source? We distribute Eucalyptus as open-source under the GPLv3 license, both as source code that will build, and as packages for the "popular" Linux distributions.  We also contribute it to the Ubuntu and Debian Linux distributions directly so that it may be distributed in this way as well.  We accept contributions to Eucalyptus from the community -- these contributions come in two primary forms.  The first is in the form of code contributions or bug fixes.

Eucalyptus is most certainly a product of its community at this stage due as much to these kinds of contributions as to the code contributions themselves.

What do you think about TIO Libre definition here: http://www.tiolibre.com/guideline/tiolibre-Libre.Definition?

I think definitions of this sort are difficult to define with any precision.  For example, if we consider the definition of "openness", I think that producing a rigorous definition like many non-technical definitions that are to be applied in a technical context, is exceptionally challenging. I admire those who are able to devote themselves to this challenge, but rather than pursue a comprehensive definition, our plan is to make our software available and to work with the community to ensure it has the widest possible impact.

Interview by Elodie Pot - writer

Open Eucalyptus- http://open.eucalyptus.com/

(c) 2010 TioNews

Follow us on Twitter: @tio_news