Happy Thanksgiving! Okay, I know not everyone celebrates Thanksgiving in November (or at all) but I think it’s a good practice to be thankful in general. Probably should do it more than one day in November (or whenever) in general.
Anyway, a lot of serverless news has happened lately. We had KubeCon which I will dive a bit deeper into which really invigorated me. If I had to go to only one tech conference a year, I would 100% choose KubeCon.
I love the energy that we get from KubeCon and I think it’s because it’s an open source conference. At its core, open source is about community. The community coming together to build cool products and share knowledge. It attracts a lot of tech practioners doing amazing things which makes my heart leap!
I love containers and this is what KubeCon is all about! And hey, it’s about serverless too (read on later). I even got to do a lightning talk about Serverless Containers. But then my laptop decided that it didn’t want to work during the demo so whatever. It happens to the best of us.
Outside of KubeCon, there were some other pieces of serverless news so let’s dive in.
Things I Learned at KubeCon North American 2024
There are going to be hundreds of newsletters focusing on the aspects of Kubernetes and AI following the most recent KubeCon so I will let those authors focus on that.
Seeing that this is a serverless newsletter, let’s focus on serverless and how it showed up at KubeCon. After all, per InfoWorld, serverless is making a comeback!
First, to level-set, KubeCon (CloudNativeCon) is an annual conference that usually does one conference in EMEA in the spring and in North America in the fall. This year actually added India and next year will add Japan. The purpose is to celebrate not just Kubernetes but the entire Cloud Native landscape as a part of the Cloud Native Computing Foundation (CNCF). It is an open source conference geared towards practioners and not a 3-day sales pitch.
There are a few CNCF projects that exist that could be considered serverless but I want to look at two in particular, Knative and CloudEvents. I had the chance to meet some of the great contributors and maintainers of Knative such as Evan Anderson and Calum Murray. It was nice to talk to them about the future state of Knative.
It was equally fun to see Coreweave mention their use of Knative when building their architecture (full talk here). I purposefully am calling this out as Coreweave is one of the most promising startups right now when it comes to Cloud Native AI at Scale. Their use of not just Kubernetes but Knative to serve their inferencing layer is promising and validates the ability to use it.
There was bunch about WASM too so let’s dive into that…
What’s new with WASM?
WebAssembly (aka WASM) is still making big news. I mentioned KubeCon and WASM was a big topic. I had the chance to listen to Jiaxiao Zhou’s KubeCon talk titled “Running WebAssembly (Wasm) Workloads Side-by-Side with Container Workloads”. I am not going to spoil it but he shows some really cool information about ways to run WASM on containerd in a Kubernetes cluster.
In the past I have talked about how one of the shortcomings of WASM was it’s lack of an orchestrator like Kubernetes. Seeing that it can potentially use k8s now, this is promising for the future. Again, I don’t want to give away too much and encourage you to watch the linked video.
Now some other WASM news, Fermyon donated SpinKube to the CNCF. What is SpinKube? Well it is a tool that simplifies building WASM-based microservices applications on top of Kubernetes.
This story dovetails nicely with the earlier one. WASM is incredibly promising but for it to work at scale, we need a way to orchestrate the microservices. There have been several projects dedicated to doing just this but I like that this one has been donated to CNCF.
Historically, the open source community has been amazing at creating tools that revolutionize tech. After all, NGINX and Apache are the most popular web servers while Unix based OSes (largely BSD and Linux) remain popular for server operating systems. Android is the most popular mobile operating system, the list goes on and on.
I think that by sharing this problem with the open source community, we will be able to create a better solution. I do think WASM will exist side-by-side with containers in the near future. It’s still early now but it’s likely to happen. I don’t think it will surpass containers in the same way that containers didn’t eliminate VMs. They will just be two runtimes that people can use.
According to The New Stack, platform engineers are embracing WASM for serverless applications. These packages are smaller and promise to be faster.
Speaking of platform engineering.
Serverless Platforming Engineering
Look, AI is takng over the world. No I don’t mean in a Terminator/SkyNet way. You just can’t have a conversation about tech without talking about AI. It is impossible to build a product strategy nowadays without including AI in that strategy. It’s here to stay.
Another concept that has really taken off in the past 2-3 years is “Platform Engineering”. It seems to be the spiritual successor of DevOps. I honestly heard next to nothing about it in 2022 and then in 2023 and 2024, I hear it everywhere.
In short, platform engineering is buildling toolchains and workflows to help developers be more efficient and allow them to self-service. Basically, it is figuring out ways to simplify the deployment process for developers. How is this different from DevOps?
DevOps, in short, is the practice of the development teams and operator teams working in tandem to deploy an application from code to production. This usually involved creating a CI/CD pipeline to streamline the build and deploy process of the software development lifecycle but also ensuring that everything is secure, workflows are continuously being improved and that both teams are working together.
According to The New Stack, platform engineers earn more than “traditional” devops engineers. I think it’s safe to say that Platform Engineering as a concept has some staying power. But hey, if the goal is to make deployment easier for developers… why not use serverless?
The challenges around platform engineering with serverless is building a fully integrated platform. In the CNCF, there are tools like the aforementioned SpinKube that could be used to simplify WASM development. Tools like Knative ARE tools that enable you to build a platform on Kubernetes to let users self-serve. But how do we connect all these pieces?
Existing tools like GitLab can help address this and Dagger looks very promising. This provides the primitives to simplify the workflows. Backstage is promising but I also think it’s early in it’s lifecycle and there are some feature gaps that I’d like to see addressed. ArgoCD on a cluster could also do some cool things.
As I see more news on platform engineering for serverless, I will be sure to share it here!
Happy Birthday Lambda (and FaaS)
I am not a FaaS fan. That should be no secret. I think it serves a great purpose BUT I don’t like how FaaS has become synonomous with “serverless”. But, I also need to give credit where credit is due. Lambda invented a space that didn’t exist before and really got the serverless compute movement going.
So with that I say….
Happy Tenth Birthday AWS Lambda!
In my blog post on the history of serverless I talk about my friend Mark and I being at SxSW in 2015 and hearing about the launch of Lambda. To see how much it shaped the market is astounding and we have to give credit to one of the OGs of cloud serverless compute.
Closing Thoughts
There were so many talks about serverless this past month with KubeCon/CloudNativeCon and just in general. We have had modern serverless for a decade now and serverless containers are growing in popularity along with WASM!
Now as people adopt platform engineering, I anticipate that we’ll see more and more serverless adoption as that would be the best way to help developers self-serve!
—Photo courtesy kaboompics.com on Pexels—