The Devs I'm Looking For
VMware has now really entered the OpenStack arena with VMware Integrated OpenStack. It's great to see more options and cleaner integration for infrastructure engineers. One thing in the VMware blog entry has me a little puzzled, and I suspect it's not VMware's issue. VMware states that they are addressing requests from customers to help deliver "OpenStack APIs to their developers". I'm not sure I believe these devs actually exist, in the sense of large numbers of them with that specific ask, in the context of running their end product. If you are such a developer, I'm looking for you, and would love to hear what you're after. Generally, I think leadership and IT teams need to pay attention to messages like this from within, really get to the bottom of the gritty details, and make sure they're addressing the right thing.
Now, devs may want to spin up some VMs for testing and the like, and OpenStack APIs could certainly be used to do that. I totally understand and support that. Waiting on blocking infrastructure requests sucks when you're on a tight product schedule. But I truly don't believe devs want to (or should, for a number of reasons) directly manage production infrastructure, or care about how many VMs need provisioning or moving, allocating or reallocating storage quota, or reconfiguring load balancers, ACLs or other network plumbing to accommodate bursts in load, outages, upgrades or myriad other forces. And, those are the things OpenStack APIs exist to do.
What do I think devs want? Devs want PaaS technologies and platforms like AWS and AppEngine, and tools like Docker. They want to write apps and have the infrastructure just "make rocket go" and scale and adjust. Transparently. This is near-100% the mechanism startups prefer, because it's the best way to achieve speed to market (which is everything) and maintain the ability to scale near-boundlessly. Early-stage startups don't have IT teams. Verbs like "consume" mean everything to most devs, not "configure", even via APIs, when it comes to infrastructure.
That said, a startup that has grown up and gets a boatload of funding or IPOs, may consider owning their own infrastructure. OpenStack's a great option to have at that point. PaaS overlay technologies like Cloud Foundry, Openshift and others provide the agile framework devs want, and can leverage OpenStack and other IaaS under the hood. IT infrastructure engineers and admins are starting to love OpenStack to help unlock the power of SDx, working with OpenFlow, OpenStack, and other programmable infrastructure technologies. Enterprises are also finally realizing that they need a lot more speed and agility, lest they be beaten by a startup or their more-agile competitors, and are embracing SDx at remarkable pace to help do this.
Overall, I'm just trying to speak for my own experience in industry working with large teams of software engineers and infrastructure engineers, as well as seeing what works and doesn't with numerous startups I've worked with via SparkLabKC, a Y Combinator modeled business accelerator. I hope that Cloud Foundry, which notably was originally created by VMware (later spun off as Pivotal), and is open-source, becomes a big part of the One Cloud vision. It's a really powerful tool, for which VMware can be genuinely proud of creating. I really believe solutions like Cloud Foundry and Openshift are the kind enterprises need to be looking at when addressing developers' needs, regardless of what's under the infrastructure hood.
Are you an application dev who really wants the capability to directly touch and configure all the production infrastructure, and all the responsibility that comes with it? If so, please shoot me a note. I'd love to hear about your use cases, and maybe learn something new.
These statements and opinions are my own, and do not necessarily reflect those of my employer, or any sane person on the planet.