On Wednesday of this week, the OpenStack foundation invited tech professionals from start-up organizations and legacy Enterprise institutions to share their experiences with the open source cloud platform. The goal of the conference was to impart to those Enterprises yet to adopt a cloud strategy the ease with which they could deploy an OpenStack architecture in their own organization.

In the third keynote panel session, moderated by Lydia Leong of Gartner, Surendra Reddy, CTO of PARC, spoke about the role of OpenStack in his own organization. PARC, utilizing an open innovation practice is, it should be noted, an independent, wholly owned subsidiary of Xerox, dating back to 2002. PARC has been on the leading edge of such endeavors as the Ethernet, laser printing and ubiquitous computing.

Reddy began by discussing his role and the role and responsibility of PARC. Primarily, PARC focuses on Big Data applications, drawing in data from submitted sensor data and running that data through a series of parallel algorithms. One of the reasons PARC settled on OpenStack when compared to other solutions like AWS, Joyent and VMWare is because, as Reddy points out, “Our researchers are not programmers. They are not DevOps guys. I don’t have any engineers to do this automation layer.” He continued, “We looked for something where, all in, we could start working without any programming effort.”

Watch the #OEForum Opening Keynote Panel Session: Part 3

Another reason Reddy cited for PARC’s deployment of OpenStack was the agility his small support team required to conduct their research. “We need lightweight containers with which we can execute without taking a lot of deployment time. I only have six people in our support team. That’s where OpenStack really helps a lot to give the tools to the researchers to focus on their primary area of research.”

PARC, on the OpenStack architecture, was able to build the entire framework that allows them, with the push of a single button, to deploy what they termed a ‘disposable Hadoop cluster.’ “I can launch a 10 terabyte Hadoop cluster…in less than two to three minutes today and run out experiments and then take it down when we are done. That’s not possible with Amazon or any other cloud service.”

While OpenStack is on track to release their Icehouse iteration in just a few months, Reddy conceded his organization is still two to three releases behind, using the Folsom release since early last year.

OpenStack ecosystem maturity : updates should be seamless

Leong was interested to know if Reddy and PARC were finding the pace of OpenStack releases to be a challenge to their organization. Reddy pulled no punches in explaining why he has decided to hold PARC back to the Folsom release. “It’s a big pain in the neck to upgrade them,” he says. “When we upgraded from Essex to Folsom, we had to re-do everything. Those transitions are really painful.” He claimed his allegiance to Folsom centered around the fact that release still maintains the capability of completing upwards of 90 percent of his workload. “It does my work so why should I go with other versions?”

This, then, led Leong to ask the obvious question, “What would be compelling enough to cause you to want to go through the upgrade cycle pain?” Reddy’s expectations for future releases echoed, I’m sure, a lot of the thoughts and concerns of Enterprise professionals in the audience.

“Zero loss upgrade. Just roll it in the new version,” he stated. “It should work seamlessly without re-doing anything. Our OpenStack base deployment is a very complex one.” Short of offering a zero downtime, zero maintenance and risk-free upgrade, Reddy claims, “I can live with the existing version. It does my job.”

Aside from the headache associated with his last upgrade, Reddy states he has been very happy with OpenStack within his organization. “We are looking at our next big step.” PARC’s goal is to work with hardware vendors in the creation of an open innovation lab aimed at democratizing data centric algorithmic research. “Our next step is to take this and spin it out to the higher level so even our researchers in other labs will have access.” The second tack to this strategy is to bring their customer base into the mix as well. “We want customers to connect their sensors and devices into this network and use PARC by utilizing our researchers to analyze their data for them. That’s our next goal for OpenStack.”

Reddy’s panel appearance was as honest an assessment of the strengths and limitations the OpenStack solution provides as could be expected from a professional in the field. Much of his insight should be considered required viewing for anyone who has yet to pull the trigger on deploying OpenStack within their own organization.

photo credit: http://www.flickr.com/photos/mamchenkov/363542398/