I recently spent the day at Serverless Days in Austin, Texas, and it was interesting for a lot of reasons. The first was that it was in a movie theater. There was no running around to different conference halls — it was just sit in your big, comfy theater chair and enjoy the show. It was nice. There was no FOMO on the other talks — everything was very simple and singularly focused. Not to mention, this was not just any movie theater — this was none other than the Alamo Drafthouse Theater, which makes all other theaters relinquish their pride, as they could never compete with the comfort, the food, or the quirkiness that is the Alamo Drafthouse.
All creature comforts aside, this conference was a developer-oriented show, which displayed some of the benefits and drawbacks of serverless. The definition of serverless, in case you were wondering, is that serverless has no definition. That’s the trouble — and perhaps someone needs to change the name. Serverless implies there are no servers, when, in fact, it’s just someone else who owns and manages the servers for you. Much like the complexity in the name, the complexity in getting it up and running leads to similar results: chaos and confusion. This conference, however, aimed to fix these complexities and try to clarify what would otherwise be a nightmare of out-of-control costs, spinning up services willy-nilly, and throwing things at the wall and seeing what sticks.
A Brief Overview of the ServerlessDays Talks
Belinda Vennam’s opening keynote was on Knative, which is an open source project aimed at managing serverless workloads on Kubernetes. Belinda explained that through primitives such as “build”, “serving”, and “eventing,” developers can rapidly deploy serverless containers, automatically scale up and down, and continuously deploy their apps — all in a streamlined workflow. Built on top of Istio, this toolset can be integrated into any cloud platform, making it easy to get up and running.
Josh Carlisle, speaking from experience, explained that sometimes serverless, with its intention to become the unequivocal solution for developers, can cause a whole host of unforeseen problems. Josh pointed out the importance of an APM (application performance monitoring) tool in order to increase visibility into what your code is actually doing and if you are optimizing your application for serverless or are just putting duct tape on the problem.
Brian LeRoux showed us a shortcut for deploying our serverless functions. His demo was a soup-to-nuts deployment of a basic OAuth-integrated web app, taking us through the entire workflow in just a matter of minutes. With the integrated staging environment, I could see the benefits and time-saving capability of Begin.
James Wickett and Karthik Gaekwad introduced us to the dangers of signing up for serverless without thinking about securing our infrastructure. “There is no infrastructure — it’s serverless,” you might say. Therein lies the problem: A lot of developers are looking the other way and assuming security is for free when it’s not.
The first thing that pops into someone’s mind when you say serverless is Lambda. But it shouldn’t because there are lots of open source alternatives out there, such as OpenFaaS. Burton Rheutan talked to us about the simplistic nature of OpenFaaS and the freedom to use your own language and package up your functions into byte-sized containers. OpenFaaS has a thriving community and over 14,000 GitHub stars.
Erik Peterson introduced us to the brand new term “FinDevOps.” This talk was an interesting change in perspective from building an application to deliver profits, to thinking about the cost of maintenance, site reliability, and the cost of doing business.
Martin Omander introduced us to different ways to break up our monolith and, as a team, how to approach the process to get everyone on-board and thinking about microservices the way they should, as well as where serverless fits in.
Finally, Chris Munns gave us a reality check and talked to us about the true value serverless brings. Serverless is not just a fantasy.
The Future of Serverless
This conference and the talks that ensued let us all know serverless (and its misleading name) aren’t going anywhere. In order for it to have a bright future, however, we have to focus on what’s behind the clouds (pun intended). The better we understand what’s going on in the background, the better off we’ll be to better understand the technology and better utilize its strengths.
If you’d like to learn about this “up-and-coming” technology and be a part of leading the next generation of developers, check out Serverless Concepts and many other serverless courses on Linux Academy.