The technology industry is one of mass temporality and today’s hottest new thing often becomes forgotten at the dawn of tomorrow. I can think back to a number of times over the past decades when a particular technology approach was deemed the ultimate game-changer, only to itself be disrupted by something shinier and newer. But while nothing ever wins, at least not long term, there are certainly those who are on the wrong side of history in this industry.
A case in point is Kubernetes, the open source container infrastructure initiative first borne out of Borg, the operating system Google uses for its own infrastructure. Kubernetes wasn’t the first attempt to make container-based approaches gain traction in the marketplace, Docker and its eponymously-named container offering was earlier to the party. But in the container orchestration war, Docker suffered a death (or at least a disability) by a thousand cuts – hubris, a case of being too early, and far too money thrown at a nascent project and team.
Contrast that with Kubernetes who, either through good management or good luck, or a combination of the two, seems to have had a golden ride with huge buy-in from the vendor community and, more importantly, great user traction. So while other related initiatives (Docker and OpenStack, for example) have somewhat capitulated when it comes to container-orchestration, Kubernetes and the Cloud Native Computing Foundation (CNCF) it spawned, have grown from strength to strength.
And if you don’t believe me, just look at some research findings from source{d}, a company enabling Machine Learning for large-scale code analysis. In a report they’re publishing today looking at the maturing world of Kubernetes, source{d} found that:
- While the number of lines of code continues to grow well over the 2 million mark, the commit velocity has been decreasing since March 2018 which implies that the project has reached a higher level of maturity and stability.
- Even though Google is the main contributor to Kubernetes by number of commits, individuals achieve a similar number. The exact number of organizations contributing is harder to measure, but the analysis shows that people from more than 600 different email domains have contributed, including major cloud providers such as Red Hat, Huawei and Microsoft.
- At its outset in 2014, the Kubernetes project had 15 programming languages, a number that quickly increased to 35 by the beginning of 2017.
- The number of API endpoints exported in the Kubernetes codebase is stabilizing at 16,000 which confirms a level of both maturity and complexity. The decrease between some releases (during 2017) might reveal a lack of backward compatibility.
Seattle bound
And so this week sees me embark on my final international trip of the year, a visit to cold and rainy Seattle where the CNCF is holding KubeCon, a celebration of all things Kubernetes. This is actually my first KubeCon (at least in person) and I’m looking forward to comparing and contrasting the vibe there with that at early OpenStack Summits and DockerCons.
My observation from afar, watching the various KubeCon events over the years, has been that KubeCon, the CNCF and its various people have managed to expertly measure the zeitgeist of the times and create events that balance contributor excitement, the drive to broaden the initiative, the various vendor tensions and the demands of the user base whom, after all, are the reason these initiatives exist.
Indeed, one parallel initiative, that has had to pivot somewhat due to the ever-increasing success of Kubernetes, is the Cloud Foundry Foundation. Cloud Foundry is a well-regarded Platform as a Service (PaaS) offering that has, at least to some extent, had its star dimmed somewhat by the emergence of Kubernetes. In an attempt to maintain relevance, Cloud Foundry has moved to embrace Kubernetes at some levels for container orchestration and the level of capitulation is shown by a survey being released today by the foundation, which shows a striking;y stringent attempt to walk a fine line between buying in to Kubernetes while maintaining a point of difference. A couple of findings to note from the survey:
- Boosted Integration of the Latest and Greatest: There is a 16 point increase year-on-year of IT decision makers who are experimenting and adopting the latest technology to build on top of existing cloud solutions.
- Big Increase in Familiarity with Cloud Technologies: Seventy-four percent of respondents can comfortably explain “PaaS” to a colleague, compared to just 63 percent who said they could in February 2016. Additionally, 49 percent can now explain “containers,” an increase from just 24 percent.
- Awareness Directly Correlates With Adoption: As comfortability has increased, only 14 percent of respondents report they are not using or evaluating a PaaS, and only 13 percent said they are not using or evaluating containers.
- Containers Prove Their Scalability: In August 2016, 51 percent of respondents were deploying between 0 and 100 containers, and only 37 percent were deploying over 100; today, the numbers have practically flipped, with 47 percent deploying more than 100 containers and only 42 percent deploying less than 100.
All interesting stuff, but somewhat confusing from an organization that accepts that, to some extent, PaaS is passé and embracing legacy application is, at the same time, a critically important requirement and a barrier to organizational success.
My approach to KubeCon
Anyway, that aside, I’m really looking forward to joining a few thousand of my best friends at KubeCon and floating around in the sea that is Kubernetes (and all things related). I’m not specifically looking to hear technology news (although some of that features below) but more I’m keen to just gauge the feeling of the show, talk to some users and contrast KubeCon, a couple of years in, with similar initiatives from history. Since history is, after all, cyclical, it will be interesting to find an answer to questions like: “Is Kubernetes getting too much startup and venture funding and hence making the early mistakes that OpenStack made?” or “With all these vendors involved, does the CNCF risk getting torn apart by the various factions.” As I said, KubeCon is going to be a fascinating end to my year.
But this is, after all, a technology event and hence there will be plenty of news around the halls. Below is a quick wrap up of what has come across my desk in the buildup to the event.
Red Hat (and friends) embrace the next cool thing with today’s cool thing
While serverless computing and containers may not be apples with apples, the fact is that, as a technology progression and, at least at some levels, serverless will be to containers what containers were to virtual machines. That is the next thing that makes the last thing look, if not tired, then at least a little ho-hum.
Given that, it is interesting to see Red Hat, Google and others collaborate on Knative, an open standard for serverless offerings. These two vendors, along with perennials SAP and IBM are giving Knative the “big tick” to deliver serverless portability across their Kubernetes installations. Red hat is tying this all in with a bunch of different things and, in something of a mind f4ck for those who don’t follow this world obsessively, Red Hat informs me that:
Red Hat OpenShift plans to add support for Knative, in dev preview early next year. This is designed to enable users to build and run serverless applications, integrating with Red Hat OpenShift Service Mesh, based on the Istio and Kiali projects. We are also using Strimzi, which makes it easier to run Apache Kafka on OpenShift or Kubernetes, via Red Hat AMQ Streams for reliable eventing and Camel-K, a lightweight integration framework built from Apache Camel that enables multiple event sources to be used as triggers for serverless applications.
Oh my, that word soup tastes mighty fine.
A cat with ninety lives, Mirantis deepens its Kubernetes offering
I love the Mirantis teams. Back when they had plenty of money to burn their parties were legendary, if with a little bit of craziness that felt a little bit like cold-war KGB honey trapping. Since those heady days, Mirantis have downsized, but remained alive, and continue to pivot like the Bolshoi’s leading lady as the winds of fortune change.
This week sees Mirantis launch the latest version of its cloud platform, including the ability to deploy Kubernetes on-premises as well as a bunch of quality assurance areas. Given Mirantis history of helping organizations deploy cutting-edge technologies, these QA aspects could be regarded as being borne in the crucible of real-world enterprise situations.
This latest version of Mirantis’ product follows on the heels of the recently announced MCP Edge, which offers operators a solution designed for edge cloud use cases. The latest release bundles up Mirantis’ bets in a number of directions and includes upgrades and updates to the DriveTrain component upgrade pipeline, granular OpenStack Ocata to Pike Upgrade pipeline, Kubernetes Upgrade Pipeline, as well as security improvements
News from… everywhere
My inbox runneth over with KubeCon news, and that is, at least at some levels, a good indication of just how much excitement there is around the CNCF. From Kaloom announcing the integration of OpenShift and Kubernetes into its Kaloom Software Defined Fabric (SDF) to Atomist, the software delivery automation company will announce new abilities around delivery to Kubernetes for free using open source Software Delivery Machine(SDM) in local mode. From customer win news like that of Platform9 who is be announcing five new enterprise customers that have selected its SaaS-Managed Kubernetes Hybrid Cloud Solution to ease-of-use announcements such that of IPO-hopefully Canonical launching MicroK8s which promises that users will be able to deploy Kubernetes in seconds. From Mesospehere’s news (remember them?) of the latest DC/OS update, including the launch of Mesosphere Kubernetes Engine to news from Crossplane, the provider of an open cloud multicloud ( how many clouds?) control plane. Crossplane makes workload portability across AWS, Microsoft Azure, Google Cloud and on-premises all from one control plane.
Suffice it to say that the week (or, in my case, two days) will be a busy one.
Back to basics
It’s easy to get led astray by all the newsy stuff when, realistically, enterprises that ask questions of me are more interested in the long-term viability of an initiative and its ecosystem than the short-term excitement of the combination of technology A, technology B and service provider C. SO that’s what I’ll be focusing on, getting a decent read on the today and potential tomorrow of Kubernetes and the CNCF. Maybe I’ll see you there?
