Cloud Native Computing Foundation (CNCF) has been active in setting up a working group for Serverless in order to define the standards. The working group has been active in defining a level of abstraction that could both be complementary and competitive to what CNCF represents. This is also an insurance policy against Serverless driving adoption away from Kubernetes. Today, they announced CloudEvents.io, a specification for describing events data in a common way.
Why should CNCF care about Serverless?
CNCF has been pushing Kubernetes as the platform for cloud native workloads and bringing together software that will enable organizations to modernize their IT. They are doing a pretty good job at building the community and it is important that they also think about the longevity of CNCF and learn from the mistakes of other OSS Foundations. Think of how OpenStack and CloudFoundry foundations went irrelevant fast because the technology landscape underneath changed quickly and the community was found sleeping on the wheels. The only way for CNCF to avoid disruption of its projects by Serverless is by taking a pro-active role in enabling the technology. Yes, Serverless is a threat to Kubernetes in my opinion.
Are they doing it right?
This is where I am a bit skeptical. I was a big believer in open source driving the standards and a big advocate of OpenStack not making their API AWS compatible (proprietary API). Having established my long held credentials about OSS, I want to put forward my thoughts about my skepticism.
- Market forces should play a role in shaping the standards. Yes, I still believe that open source is critical to developing standards but I think listening to the market is important in keeping the standard relevant. If OpenStack is a lesson, we cannot (and should not) ignore the market leader(s)
- Talking about open standards without staking out on a single project doesn’t lead to anywhere. Kubernetes is the leader in Cloud Native world but, in the world of On-Premises Serverless platforms, Apache OpenWhisk is gaining traction with one public FaaS already built using the project. Putting together a working group advocating standards without staking out the foundation on a public FaaS or an On-Premises Serverless platform, will have very little impact. Remember TOSCA?
In my opinion, I really like what CNCF Serverless Working Group has been doing but I want CNCF to pick one of the many Serverless projects as the foundation for this working group. Plus, if they get folks from Amazon AWS and Microsoft Azure involved in the process, there will be more legitimacy. Without this, this is yet another standards attempt that will go nowhere. Thoughts?
On-Premises Serverless Ecosystem: Apache OpenWhisk, Kubeless, Nuclio, Project Riff, OpenFaaS, Oracle Fn, etc. See the complete list on Github.
Here is an unscientific poll I ran on Twitter
With no single dominant On-Premises Serverless (@#$*&) offering in sight and On-Premises adoption still low, CNCF seems to be going overboard in their strategy. Is it a reasonable analysis? Love to hear your thoughts
— Krish Subramanian (@krishnan) May 3, 2018
Update: Chris Aniszczyk from CNCF says AWS is being represented in the working group calls. We will follow AWS contribution in the following months, I just hope this effort results in true interoperability and not yet another effort that didn’t help.