Serverless computing can save your company money and maintenance time, among other things. Tom Merritt lists five things to know about serverless computing.
Server room? You don’t have room for that? You’re not doing peer-to-peer here–you still need a server, but you don’t need it all the time. All of these seemingly contradictory considerations are no problem when you take advantage of functions as a service (FaaS). Here are five things to know about serverless computing.
SEE: Server migration checklist (TechRepublic Premium)
- It doesn’t mean no server. It just means you don’t have the server the whole time. Sometimes people confuse this with decentralized protocols like peer-to-peer. You’re just not using the server all the time.
- It can save you money. Instead of paying for servers even when you don’t need them, you can pay based on the time and memory allocated to run your code.
- You save on maintenance. Since you’re not managing a server you’re not managing an OS, which means you’re not managing the patches and maintenance that go with it. That of course can be a downside since all of that is under control of a third party service.
- It’s not all rosy. Performance may degrade because code may spin down when not in use. You also can’t attach profilers, debuggers and such. Serverless computing also isn’t good for high performance computing.
- The serverless supercomputer. In a Stanford seminar, Keith Winstein posited what would happen if you took an app that would normally take one hour to run and spun up 3600 AWS lambdas that each run for one second. He posited you’d get near instantaneous results. Spinning up 3600 machines, even virtual ones, would make this impractical. Serverless compute makes it actually possible.
Whether it’s AWS Lambda, Google Cloud Functions, IBM Cloud Functions, or Azure Functions. Serverless compute is another good tool to consider. Depending on your needs, you may be better off ditching the server, and calling functions as a service.