I posted this to Twitter:
⚡ I need your help! #serverless poll!
I'm curious 🦝
In your opinion, what is the most important feature that MUST exist for a service to be defined as "serverless" as opposed to "fully managed"? Please RT for reach. If your option doesn't exist, please explain in a reply!21:46 PM - 12 Jun 2019
If the poll is still active when you read this, feel free to vote and share with your network.
I'm very curious. There are a lot of cloud services described as serverless but not a single, unanimously accepted industry definition. A few common ideas include:
- Micro-billing (pay-per-use) - no charge if you're not actively using it
- Event-driven/code - small code footprint that is triggered by events
- Auto-scale - automatically scales up or down on demand
- No configuration/setup - no need to set up a VM or configure a cluster
Some examples include Azure Functions, AWS Lambda, Google Cloud Functions, OpenFaaS, and Apache OpenWhisk.
I'm curious, based on your knowledge, what do these services have in common that makes them special? What should qualify a service as "serverless" as opposed to the "fully managed" options we've had available for some time?
(There is no right or wrong answer here, trying to understand the general views on this).
Top comments (1)
For me, it's mostly about having to think (less) about servers. As in, I'm managing the mental overhead of one less layer. The four bullet points you list are techniques that vendors use in order to facilitate this:
cron
, etc.The flip side, to me at least, we trade servers for services. By becoming serverless, we also tend to being servicefull. It's a good trade, in my opinion, because it shifts the focus away from managing infrastructure and onto tools that will help you achieve your objectives.
I suspect, however, that we (the community) will continue to have these discussions for a very long time!