I’d prefer my toilet to not fail fast – #InnoIT Think Tank

Dell World Social Think Tank - Enabling Innovation in IT.

Earlier this week at Dell World I sat in on an afternoon Think Tank moderated by TechCrunch’s Alex Williams. Essentially, we discussed the challenging role of IT now-a-days. Per usual, there was much discussion of getting IT to be more innovative and the “threat” that new IT delivery methods like cloud and consumer technologies bring to the status quo. Because technology can do so much, so much faster now-a-days the IT department has a huge challenge and a contradictory mission: IT has to keep the lights on, be stable, and at the same time innovate their brains out.

Dell World Social Think Tank - Enabling Innovation in IT.

Being a professional observer of the IT industry and its history I’ve often found that those two things require different processes, different people, and different technologies. The mind set of keeping things stable a reliable (the five nines crowd) doesn’t fit with coming up with new stuff. Practices like Agile and the rapid delivery cycles in DevOps can help, but at some point, the two paths of ensuring stability and profiting from disruption are divergent enough that you can’t perfectly co-mingle them…and yet, that’s what we expect from the IT department.

I’ve been reading Taleb’s latest book, Antifragile and I’m really liking the premise of it: you want to build systems that benefit from failure and disruption. There might be something of a middle-ground in that nuance, and it’s certainly a way of thinking that cloud has benefited from. We’ll see how quickly we can get IT – and corporate! – culture to start embracing failure as education and helpful instead of something to be avoided even to the point of doing nothing instead of trying.

Project RIPTide – the second Dell incubation project

Among my other doings at Dell, I helped put together and now run the internal incubation program. The idea of the program is to provide a sort of internal angel investment fund and program for employees who want to develop ideas. Project Sputnik is the first project we’ve done, and the second one, RIPTide is now emerging into public.

The idea behind Project RIPTide (headed up Shree Dandekar) is to pull together a business intelligence platform for the mainstream. There’s a tremendous amount of now affordable horse-power (in hardware and software) available to customers of any size now but the last mile of hooking it up all together is a tough one. Shree’s idea was to take advantage of Dell Boomi, some beefy hardware, and BI tools to help smaller companies get better analytics over their business.

When we were in the concept phase of this project, I was cooking up all sorts of fanciful use cases like helping food-truck owners analyze Twitter in real-time to find where people are longing for a taco…so they can drive the truck over there to sell some grub.

It’s been exciting to watch Shree put this project together and even find an initial customer, Team Express. There’s more to go with the project and we’d love to get your input if you’re interested in this kind of thing. Check out Shree’s overview from earlier this week.

Sputnik Cloud Launcher – Doing More DevOps

One of the tools in Project Sputnik is the “cloud launcher.” The idea for this tool is to help instrument a DevOps life-cycle: the tool models out a simulated cloud on your desktop during development, and then deploys it to “real” clouds once you’re ready. We demonstrated one version of the cloud launcher at Dell World this week that uses juju.

In the meantime, OpsCode’s Matt Ray has been working on another approach (which he describes in the above video) that uses Chef under the covers. See the code checked into the Sputnik repo as well. I’m looking at these two versions as proofs of concept, or even “spikes” to explore how to best implement the idea. We’re eager to get feedback and engagement from the community to figure out which approach (or a third!) is most helpful.

Project Fast PaaS and Dell Cloud Labs

A couple of developers in our Dublin cloud labs started working on Cloud Foundry and set it up to run on our Dell cloud. You can check out more info and sign up for a invite to it.

Moving Beyond The PaaS Paradox

In my strategy role I’ve been looking at PaaS for awhile now. In doing that, I keep hitting upon what I call “The PaaS Paradox.” If you take any given analysts forecasts for PaaS, the overall market looks “bad” compared to IaaS and SaaS: $2.9B by 2016 by a recent Gartner estimate – or about 3% of the ~$110B public cloud market in 2016 (I subtracted out that annoying “advertising” segment that Gartner tracks).

And then you have some real gorillas already moving in there: Microsoft, Salesforce, Google, IBM, Oracle, and so on. While several billion may seem amazing to individuals, in the IT industry, it’s not much…esp. if you’re competing with those guys. (As another data point along the PaaS road: EngineYard helpfully reports its revenue from time-to-time, $28M back in July, 2011.)

And yet, everyone is always going on about how PaaS is mega important. Each year it’s going to be “the year of PaaS,” and analyst survey data always indicates high interest in PaaS.

My theory has been that when most people, esp. all those gleeful survey respondents, think of PaaS they’re not thinking of “pure PaaS” (or 1st and even 2nd generation PaaS). Instead, they just are thinking “doing software development with cloud technologies and practices.” Once you re-calibrate your whiz-bang charts to include all of software development, “PaaS” seems a lot more attractive.

I ran this by Jeffery Hammond and James Staten in a conversation the other day and they framed it in another, interesting way: people want the ability to run, and target different frameworks in a cloud context. Heroku is the classic of example of this. While Heroku is a PaaS, it’s more about being able to run rails (and plenty of other languages and frameworks now). This flexiblity fixes that unsettling feeling that 1st generation PaaS had: you were using, essentially, a propriety framework that was limiting your choice.

Or, as Stephen puts it: PaaS is the new middleware.

With that framing, you can escape the PaaS Paradox, and PaaS is a lot more interesting. So far, Cloud Foundry has seemed one of the better architectural fits for this “PaaS as middleware” think.” As we move “Project Fast” through (the new) Dell Cloud Labs, I’ll be eager to see how that architecture plays out and even more excited to see how the Dell community reacts to and participates in the project. As with Project Sputnik, a huge part of what we’re doing is engaging with developers, which sounds like a pretty good way to spend time to me.

Also: check out some demo videos of Project Fast PaaS.