Tuesday, October 26, 2010

Using cloud to test deployment scenarios you didn't think you could

Often, IT considers cloud computing for things you were doing anyway, with the hope of doing them much cheaper. Or, more likely from what I’ve heard, much faster. But last week a customer reminded me about one of the more important implications of cloud: you can do things you really wouldn’t have done otherwise.

And what a big, positive benefit for your IT operations that can be.

A customer’s real, live cloud computing experience at Gartner Symposium

The occasion was last week’s Gartner’s Symposium and ITXpo conference in Orlando. I sat in on the CA Technologies session that featured Adam Famularo, the general manager of our cloud business (and my boss), and David Guthrie, CTO of Premiere Global Services, Inc. (PGi). It probably isn’t a surprise that Guthrie’s real-world experiences with cloud computing provided some really interesting tidbits of advice.

Guthrie talked about how PGi, a $600 million real-time collaboration business that provides audio & web conferencing, has embraced cloud computing (you’ve used their service if you’ve ever heard “Welcome to Ready Conference” and other similar conferencing and collaboration system greetings).

What kicked off cloud computing for PGi? Frustration. Guthrie told the PGi story like this: a developer was frustrated with the way PGi previously went about deploying business services. From his point of view, it was way too time-consuming: each new service required a procurement process, the set-up of new servers at a co-location facility, and the like. That developer tracked down 3Tera AppLogic (end of sales pitch) and PGi began to put it to use as a deployment platform for cloud services.

What does PGi use cloud computing for? Well, everything that’s customer facing. “All the services that we deliver for our customers are in the cloud,” said Guthrie. That means they use a cloud infrastructure for their audio and web collaboration meeting services. They use it for their pgi.com web site, sales gateways, and customer-specific portals as well.

Guthrie stressed the benefits of speed that come from cloud computing. “It’s all about getting technology to our customers faster,” said Guthrie. “Ramp time is one of the biggest benefits. It’s about delivering our services in a more effective way – not just from the cost perspective, but also from the time perspective.”

Cloud computing changes application development and deployment. Cloud, Guthrie noted in his presentation, “really changed out entire app dev cycle. We now have developers closer to how things are being deployed.” He said it helped fix consistency problems between applications across dev, QA, and production.

During his talk, Guthrie pointed to some more fundamental differences they've been seeing, too. He described how the cloud-based approach has been helping his team be more effective. “I’ve definitely saved money by getting things to market quicker,” Guthrie said.

But, more importantly, said Guthrie, “it makes you make better applications.” “I was also able to add redundancy where I previously wouldn’t have considered it” thanks to both cost efficiencies and the ease with which cloud infrastructure (at least, in his CA 3Tera AppLogic-based environment) can be built up and torn down.

“You can test scenarios that in the past were really impractical,” said Guthrie. He cited doing side-by-side testing of ideas and configurations that never would have been possible before to see what makes the most sense, rolling out the one that worked best, and discarding those that didn't. Except in this case, "discarding" doesn't mean tossing out dedicated silos of hardware (and hard-wired) infrastructure that you had to manually set up just for this purpose.

Practical advice, please: what apps are cloud-ready?

At the Gartner session, audience members were very curious about the practicalities of using cloud computing, asking Guthrie to make some generalizations from his experience on everything from what kind of applications work best in a cloud environment, to how he won over his IT operations staff. (I don’t think last week’s audience is unique in wanting to learn some of these things.)

“Some apps aren’t ready for this kind of infrastructure,” said Guthrie, while “some are perfectly set for this.” And, unlike the assumption many in the room expressed, it’s not generally true that packaged apps are more appropriate to bring to a cloud environment than home-grown applications.

Guthrie’s take? You should decide which apps to take to the cloud not by whether they are packaged or custom, but with a different set of criteria. For example, stateless applications that live on the web are perfect for this kind of approach, he believes.

Dealing with skepticism about cloud computing from IT operations

One of the more interesting questions brought out what Guthrie learned about getting the buy-in and support of the operations team at PGi. “I don’t want you to think I’ve deployed these without a lot of resistance. I don’t want to act like this was just easy. There were people that really resisted it,” he said.

Earning the buy-in of his IT operations people required training for the team, time, and even a quite a bit of evangelism, according to Guthrie. He also laid down an edict at one point: for everything new, he required that “you have to tell me why it shouldn’t go on the cloud infrastructure.” What seemed draconian at first, actually turned out to be the right strategic choice.

“As the ops team became familiar [with cloud computing], they began to embrace it,” Guthrie said.

Have you had similar real-world experiences with the cloud? Or even contradictory ones? I’m eager to hear your thoughts on PGi’s approach and advice (as you might guess, I much prefer real-world discussions to theoretical ones any day of the week). Leave comments here or contact me directly; I’ll definitely continue this discussion in future posts.

No comments: