Last week was another installment in the ServerlessConf series, beginning last May in Brooklyn, New York and in November in Tokyo. For a 3-time conference, Pete and the Cloud Guru team have been champs at keeping everything organized and bringing in great speakers. This time, they found a venue in the City of London, and packed it with 400+ attendees for two days.
Day 1 Vlog
It’s not a real ServerlessConf without some temperature problems. ServerlessConf NYC was literally the hottest conference: temperatures in the venue climbed to 90+ at some points. This time, it was the fire alarm. Right in the middle of setup for day one everyone was kicked out for an hour or so.
Before I get into talks, there were 40+ speakers at the event. If I don’t mention your talk here it’s not because I didn’t enjoy your talk – two tracks mean I could only cover half of them at most. With that disclaimer, here’s a few of the talks I was able to catch that were highlights for me.
Cloud Custodian (Rules Engine) & Cloud Detour (Chaos Engineering tool)
The team from Capital One (a ServerlessConf sponsor) had two talks. The first was on their continued development of their open-source Cloud Custodian tool that provides automated remediation and alerting for AWS resources. Of course, automating policies like tagging, billing categories, and security groups become critical when using AWS at their scale and number of engineers. Drew Firment, Capital One’s Director of Cloud Engineering, has a blog post going into much more detail if this catches your interest.
Another team from Capital One showed off Cloud Detour, which is a “Serverless Chaos Engineering” tool. Effectively, it performs functions similar to Chaos Monkey. The difference is that it runs from AWS Lambda to inject failures into your running applications. It also has features to let teams approve failure plans and write their own failure scenarios.
Jared Short from Trek10 called a last-minute change on his talk, skipping his operations talk in favor of leading a session using Lambda and API Gateway to provide GraphQL APIs to React (and other) frontends. The isolation and scalability provided by Lambda makes it easy to fan out GraphQL resolvers and keep users from affecting each others’ performance and availability. It also means different teams can provide resolvers and all implement them in separate Lambdas.
The Making of Azure Functions
Chris Anderson, @crandycodes from Microsoft showed off new features in Azure Functions. The big story here is that the open-source Azure Functions runtime has enabled them to publish awesome debugging tools. In my opinion, they should be shouting from the rooftops about their runtime being open and runnable locally for testing. Chris showed off using the Visual Studio debugger to step through functions in exactly the environment that they’d be running in.
If needed, you can even connect with a remote debugger to live functions in Azure so it’s a true production environment. Keep an eye on this, because developer experience is one of the oft-cited advantages of “serverless” and making debugging this nice is a big deal.
Serverless architecture at iRobot
iRobot’s Ben Kehoe used his session to cover architecture, latency profiles, and the future of deploy frameworks. Roomba (iRobot’s consumer robo-vacuum) has an iPhone app for controlling scheduling and other functions, and the backend for that system is serverless. With AWS IoT, API Gateway, Lambda, and DynamoDB they provide an API for the app, a controlplane for devices, and a low-cost way for iRobot to continue making their products smarter.
He also brought a (serverless) Roomba 900-series to raffle off.
Why PubNub moved Serverless Computing into the Network
PubNub showed off their relatively recent entrant into the serverless/Function-as-a-Service market: PubNub BLOCKS. PubNub is a low-latency global data network, and they’re known for providing messaging systems for IoT, mobile, and realtime web applications. BLOCKS is a way to extend your compute footprint into their network, so you can have parts of your application living very close to your consumers that have access to your control plane.
The best way to describe it is as CDN, but for compute. They already have PoPs all over the world, and BLOCKS comes with a catalog that has pre-written functions for geolocation, image manipulation, and the IBM Watson series of APIs.
Globally Available Serverless Architectures
Also in the “going global” theme, Rich Jones showed off new features to Zappa and talked about expanding to multi-region Lambda deployments. That gets you closer to the consumer, makes you more resilient, and can be important for regulatory compliance. I interviewed Rich a while back about Zappa and using existing web frameworks with Lambda.
OpenWhisk Under the Hood
The OpenWhisk team at IBM showed more about the internal architecture of OpenWhisk, and talked about how it handles cold/warm/hot container starts when users are allowed to provide custom containers instead of being locked to a couple language runtimes. Like the Azure team, they had great new debugging tools to show off and it’s interesting that the open platforms so far seem to be way ahead on platform visibility and debugger support.
In part, that’s probably a function of them not being worried about giving away any secret sauce since you can download their whole platform for free (and they want you to!). IBM also talked a bit about their upcoming API gateway system built on nginx to link up HTTP requests as an OpenWhisk trigger.
Serverless for Developers: Tips for Your Next App
Danilo Poccia talked about the developer’s perspective of serverless architectures, which ties in with his soon-to-be-released AWS Lambda in Action book. My biggest takeaway from this talk was that when moving to FaaS/serverless when you also have an existing legacy system, the same techniques for microservices migrations are applicable. Serverless for Developers slides are now available.
Ansible for Serverless
I also gave a talk on using Ansible for serverless infrastructure and on using it with the Serverless Framework. You can pick up my slides here, and the recordings for all the talks will be available soon.
Mike was up late editing another vlog for day 2, sacrificing his forty winks (which is what Brits call sleep, as I understand it) for YouTube fame and fortune.
As the conf went on Mike and others spent some time taking video and sacrificed sleep to get them edited.
Day 2 Vlog
With that last vlog, that’s it for this recap too! Hope to see everyone (plus new faces) at ServerlessConf $NEXT! Some options for the next venue are San Francisco or Seattle. Both have solid name themes too: I Left my Servers in San Francisco or Serverless in Seattle.