The email sent will contain a link to this article, the article title, and an article excerpt (if available). For security reasons, your IP address will also be included in the sent email.
When Adrian Cockroft—the guy who helped put the loud in Cloud through his energetic evangelism of Cloud Native and Microservice architectures—talks about what’s next, it pays to listen. And you can listen, here’s a fascinating forward looking talk he gave at microXchg 2017: Shrinking Microservices to Functions. It’s typically Cockroftian: understated, thoughtful, and full of insight drawn from experience.
Adrian makes a compelling case that the same technology drivers, faster networking and cheaper messaging, that drove the move to Microservices are now driving the move to Functions.
The payoffs are all those you’ve no doubt heard about Serverless for some time, but Adrian develops them in an interesting way. He traces how architectures have evolved over time. Take a look at my gloss of his talk for more details.
What’s next after Functions? Adrian talks about pushing Lambda functions to the edge. A topic I’m excited about and have been interested in for sometime, though I didn’t quite see it playing out like this.
Datacenters disappear. Functions are not running in an AWS region anymore, code is placed near the customer using a CDN at CDN endpoints. Now you have a fully distributed, at the edge, low latency, milliseconds from the customer way of running code. Now you can build architectures that are partly in the datacenter, partly at the edge, and partly at the customer premises. And since this is AWS, it’s all, of course, built around Lambda. AWS Greengrass and Snowball Edge are peeks into what the future might look like.
There’s a hidden tension here. Once you put code at the edge you violate two of Lambda’s key assumptions: functions are composed using scalable backend services; low latency messaging. The edge will have a high latency path back to services in the datacenter, so how do you make a function based distributed application at the edge? Does edge computing argue for a more retro architecture with fewer messages back to a more monolithic core?
Or does edge computing require something completely different? Here’s one thought as to what that something completely different might look like: Datanet: A New CRDT Database That Let's You Do Bad Bad Things To Distributed Data.
Now, let’s see the future by first taking a tour of the past….
From Monoliths, to Microservices, to Functions
Return to Article