I’ve just returned from OpenStack Summit 2015 in Vancouver where 6000+ software-defined attendees convened to contribute and commune and where the opening keynote speaker laid down the following proclamation:
Every company is becoming a software company and every company has to write software to compete.
Thus began a conference where presenters representing enterprise IT shops like BestBuy, TD Bank, and Walmart as well as vendors showed how to turn racks of commodity parts into large scale, multi-application compute clouds with the transformational power of software. And yes, here the basic foundation of computing is still built on CPU, storage and networking gear. But one of the points of the OpenStack movement is to render the machine room stuff free from vendor lock-in so that enterprise users and software developers (often one and the same in the OpenStack world) can do big compute cloud things without a paying a big bill.
This was admittedly my first OpenStack Summit. Some I spoke to there were veterans. But even so I felt I got a good enough feel for the current state of OpenStack to test some common perceptions that some within the OpenStack community would call misperceptions:
The organizers hosted a private session for industry analysts. Five of the presentations we saw were from enterprise IT users, but because most of them were given under non-disclosure, I can’t mention names. I can say however that only one fell into the biggest-of-the-big category. Furthermore, the number of people involved with these projects highlighted in the analyst session was small. A large financial services firm is doing an amazing amount of OpenStack in production with just three people.
Here’s the rub: Finding these people is admittedly hard at the moment because OpenStack does require coding and the skill set is specialized. There were help wanted signs tacked to booths all over the show floor. In fact more than once I heard someone mention that there was more recruiting on the show floor than selling. I overheard someone also mention that they had paid to turn one of their IT admins into an OpenStacker, only to see him take a better offer from a competitor. But with the right people, a lot can be done with a few. Automation is key. Documenting the environment will be key as well lest the knowhow walks out someday.
Big vendors are undoubtedly behind OpenStack, but actively pushing it forward? Maybe not so much. If OpenStack were to die tomorrow, would these vendors resurrect it? I don’t think so and the reason is simple. The OpenStack mavens believe they can white-box an entire data center by pitching those greedy vendor’s logos onto the scrap heap. One user demonstrated a way to get the cost of enterprise array storage down to less than $0.10 a GB. Can big storage vendors even buy from ODMs at that price? Furthermore, most of them already have private cloud solutions to offer. Why would they need another one?
OpenStack is a user-driven movement. It can and likely will succeed because a growing, world-wide user community is pushing it forward. OpenStack is now seen by some CIOs as a place to land all new applications and a future home for their existing ones. Big vendors are supporting OpenStack as a defensive move. They’re getting out in front of the wave to ride it because they fear it will wash over them if they don’t.
It is true that science projects like to have PhDs running them. A main tent presenter’s comment that OpenStack shouldn’t require a PhD to run it drew applause. So, point taken. It is also believed as an article of faith among analysts that OpenStack has only penetrated 5% of enterprise IT, if that.
However, while the percentage of OpenStack implementations out there are actually “in production,” vs. waiting for a use case is hotly debated, certainly one of the points of the summit was to showcase OpenStack doing real enterprise production IT. The implementations analysts were shown by users for example were indeed hosting production applications. There are people out there that know how to get beyond the science experiment phase, and the number is growing.
It is also true that code-base stability is an issue. The OpenStack community is aware and has begun to address the issue with a project called DefCore which defines the most stable releases of the various projects to include with its standard distribution and certify it against vendor’s use of the OpenStack logo.
(Note to the big vendors involved with OpenStack on the subject of code base stability: Some of you have taken to proclaiming that you contribute tens of thousands of lines of code every month to OpenStack as a way to demonstrate fervent support and one-upping your competition. Does that aid or detract from the perception of stability?)
In fact, the list of use cases is surprisingly long as was demonstrated at the Summit. But more importantly, their implications are huge. One of the things OpenStack is absolutely about is large scale enterprise cloud computing at low cost. So if you’re a CIO or CFO and the bill from your favorite public cloud vendor is shooting through the roof, it might be time to get the OpenStackers to build you one of your own. If analytics, mobile, and the Internet of Things applications you’d like to do is going to cost a bundle for adding the needed hardware and software support to your current IT environment, look at standing them up the OpenStack way. And, if you want to test the growing number of new technologies coming at you like water from a fire hose, OpenStack makes a great sandbox.
OpenStack is an IT phenomenon like none we’ve yet seen. It is big scale, diverse, and even diffuse. And it has captured the focus of a growing list of enterprise IT organizations because its messages are powerful and its implications are far reaching. My post-Summit sense however is that if the expected maturation process takes too long, momentum will be lost. The community needs to keep the momentum going that was demonstrated at the Summit. If it does, open source will profoundly impact IT once again.