When you look back at decades of computing history, the arc bends toward abstraction: the process of making it easier and easier for people to put their imagination into code by removing the complexity of actually getting software to run on hardware. Serverless computing is shaping up as the next big breakthrough in this ongoing process, and different visions of how it will actually work are starting to emerge among the cloud computing set.
We hosted five of those thinkers at the recent GeekWire Cloud Tech Summit in Bellevue, Wash., devoting one of our five technical tracks to serverless computing. Serverless computing centers around the idea that software developers shouldn’t have to know anything about the hardware on which their code will run; there are still plenty of servers, but serverless services (say that ten times fast) allow developers to build and deploy apps without having to manage the computing resources required for those apps.
This has a number of cool effects, such as allowing per-second billing for computing resources and enabling low-power computing devices to run more sophisticated applications built around events, triggers, and functions. It’s particularly attractive to developers building applications that automate IT infrastructure or enable the internet of things, and it is attracting a great deal of interest among the developers hoping to skate to where the puck will be, as Wayne Gretzky would have put it were he a cloud-native developer.
And as happens with technologies on the horizon that offer a lot of promise once the kinks are worked out, competing factions are starting to coalesce. Amazon Web Services’ Lambda technology introduced in 2014 basically created the market for this type of service, but Microsoft and Google have thrown a ton of resources at serverless computing in hopes of catching up.
Here’s how five experts presented the future of serverless computing during the 2018 Cloud Tech Summit.
Chase Douglas, co-founder and chief technology officer, Stackery
Douglas kicked off the day’s serverless sessions with a reality check about the promise of serverless computing and what actually needs to happen to take advantage of this technology. A lot of the monitoring and testing tools needed to make sure serverless code will be stable in a production environment are immature, and developers that want to use this technology need to go into it with eyes wide open.
Aparna Sinha, product management lead, Google
Sinha gave Cloud Tech Summit attendees a sort-of preview of what Google planned to announce almost exactly a month later at its own Cloud Next 2018 conference, where it introduced the notion of running serverless code on a new project called knative, which involves deploying that code to Kubernetes clusters. This idea is very much in keeping with Google’s instincts to make Kubernetes a huge part of its cloud product strategy, and Sinha walked through the benefits of such an approach, which mainly center around portability.
Donna Malayeri, product and community manager at Pulumi
Inevitable battlelines are being drawn around serverless computing and containers, two of the cutting-edge technologies in discussion among forward-thinking development shops. But they don’t have to exist in separate worlds, according to Malayeri, and Pulumi is working on technology that will support any kind of deployment strategy that developers want to use for their code, be it containers or serverless tech.
Matt Weagle, director of infrastructure, ShiftLeft
How does serverless computing work with the concept of microservices, which is a huge shift in application-development thinking that is well underway? Weagle walked attendees through ways that serverless computing services like AWS Lambda can be used in concert with applications built around microservices.
Sanath Kumar Ramesh, software development engineer II, Amazon Web Services
Ramesh took attendees through the AWS Serverless Application Model, which allows developers to specify how their applications should be deployed without having to write as much code as might be required on their own. As tools built specifically for serverless development mature, it seems like there will be an equal movement on the part of serverless providers to build in support for existing tools that developers already know and love.
