Coté

While this dude’s tone is pretty harsh, there’s not too much wrong here if you peel that back. The issue is one of contextualizing OpenStack. I think a lot of people want it to be a finished, done product.

There’s even the sense from some OpenStack die-hards I’ve spoken with over recent years that commercializing at this point is a joke: it’s too early. I don’t think any of them realize that’s what they’re thinking, but when you hear about modularity and customization, it’s a good sign that the person is implying “it’s not fully baked yet” (though not always).

The rest of the world is desperate for OpenStack to be a full fledged product now. Something that you could download and get up and running quickly in the same way that I could download Windows Server and get it up and running after installing it. I don’t think the community really wants that at the moment and is instead looking at vendors to do that over the coming years. 

That message never gets articulated well. The OpenStack community is very developer-think oriented, not product oriented. What it needs is some additional product marketing efforts to explain itself. The Eclipse Foundation is very good at this: “The Market” is rarely confused about how Eclipse projects fit in and expectations are generally set. And at the same time, within the Eclipse community, commercialization strategies abound. TaskTop is a good example here.

Now, there’s cases of large companies working on internal private clouds with OpenStack (to me, that’s the short and medium term commercial driver of OpenStack), and this week at the IBM software analyst event I heard that many of their new SaaS offerings are running on OpenStack. We covered some of the newer case studies in a recent 451 write-up as well, and also check out our general overview of the recent OpenStack Summit (subscription required, or a 30 day trial, as always)

All of this is following the pattern of Linux or, worse for throttling the freaking out, Java. Java was and is a similarly loosely coupled chain of ideas, APIs, and code injected with all sorts of concerns from competing and conflicting vendors and “community members.” You can’t really go download “Java” and be up and running, you have to choose the JVM (Sun/Oracle, IBM, or the unofficial versions out there)…or worse for clearing things up, you download a JVM and then end up running JavaScript in it as your programming language! That weird example aside, once you pick your JVM and install it, now you need to pick an application server or some kind of runtime container from Spring to JBoss to raw servlets to whatever is the flavor of the year container wise.

There is no “Java product” to be downloaded: there’s just several engines to pick from and then a whole bunch of other stuff you need to wrap around it to finally be able to develop and then deploy an application.

OpenStack is sort-of-kinda in that same spot…except everyone is kind of confused about it. As the title here says, there are massive expectation mis-alignments in the OpenStack world and it’s clearly time to fix them.

(Geoff Arnold has an excellent write-up of this topic, you should go read it.)