It's my belief: dev, DevOps, agile and continuous development, all of these will never end in the cloud. No app will ever be finished, only tidy release numbers for comparison sake. This is awful, and this is great. The land grab is on. Containers are the new crack, dimming bright minds, one colleague tells me, while another has become fascinated at the incredibly diverse number of appliances and appliance constructs that have emerged, even in this calendar quarter.
The awful part of never-ending development means: Is that thing done? Finished? Completed? Invoice paid? Game over? In the bag? Done? Buh-bye?
We all know what the answer is: no. And that's a good thing.
Yesterday, CoreOS, friend/enemy of Docker, announced support for Tectonic, a mash-up of Google's Kubernetes platform + CoreOS. Many other small moving parts will attempt to make a machine. Not a Driverless Google Machine, but rather, a Fifth Army of containers.
It's an example of a wave of container developments designed to shake up how continuous app delivery can be rationally managed (we hope) to not only keep a serious edge on the competition, but also to stanch the mania of things like malware released before an anti-virus signature is available, wicked inter-version code incompatibilities, and internecine warfare that's driving increasing container quality and security surrounding them.
I sent a message to CoreOS. What CoreOS containerizing efforts of rkt/Rocket (an ostensible Docker competitor) in the face of joining Tectonic? Doesn't this cause dilution of efforts? Brian Philips, CTO of CoreOS, said:
"Tectonic takes advantage of the best parts of Docker, rkt, and the rest of the combined CoreOS and Kubernetes portfolio. Inside of Tectonic, rkt plays a couple of important roles. First, many of the major components of Tectonic are built as App Container Images (ACIs) and are verified and ran with rkt. Secondly, we are contributing code to Kubernetes to support the App Container Spec so a user can run ACIs natively via the Kubernetes API. So, in this way the effort we put into rkt is complementary to our Kubernetes and Tectonic work."
It takes a lot of water to lift all of these yachts, but there is energy and the kerosene of venture capital behind Tectonic.
Famously, Alex Polvi, co-founder and CEO of CoreOS, is in the thick of the discussion, citing his perceived deficiencies in Docker, and aligning CoreOS with Google via Tectonic, as well as Mesos(phere).
In the other corner of the boxing ring is gargantuan contestant Red Hat, whose Atomic Linux Instance (nice name, different era), will rival CoreOS. Together, they'll be in the Final Four along with Canonical's Ubuntu, whose ubiquitous server and stripped images seem to dominate Docker images, in terms of probably-deployed.
Should VMware quake? Microsoft?
Combine all of this with the robotry of OpenStack and its rivalry. Add in software-defined networks, each with a network definition syntax "better" than the last. Stir. Add wars and debates about suitable security at all elements of each stack, infrastructure definition, and bake with the aforementioned containerization. Take several sips of wine. Better still, several shots of the hard stuff.