I can't speak for Erlang, but I use Elixir and not primarily for the concurrency model. I just really like the language (and it's functional-but-accessible approach), relative speed (compared to Ruby), fault-tolerance, community.
In particular the fault tolerance and process-based approach is just awesome. I regularly find myself writing code with just the 'happy path' (and only the obvious error handling), which just feels a ton nicer than the more defensive approach I generally take in other languages.
The tooling around the language is also really nice.
Erlang has way more advantages than its concurrency model but none of these can be used in a lambda I guess. It is just good to have diversity, many will disagree but you get used to love the Erlang syntax and being able to run it in AWS lambda is kind of nice to not change the stack.
I wrote a Lambda function that does lots of concurrent checks and then reports back the results at the end of its execution. Concurrency works fine for my usecase.
I've been saying for a while that Lambda (and serverless in general) should just be a compilation target, not a Radically New Computing Concept(tm). This is another inch towards that.
I think it is already through libraries. At least for lambda in python you have libraries (Zappa for example) that allow you to write basic flask and Django apps and then deploy them to lambda. These apps can still be deployed the "old" way with no or minor modifications
Does this run a separate Beam instance for every Function? If so, isnt that an enormous waste of Erlang? If not, how isolated are requests? Can I send messages between them? Should I?
Lambda will create a function instance to handle concurrent requests, i.e. if 100 requests happen at the same time there will be 100 function instances. It will then however keep them "alive" for a few minutes allowing it to reuse already running instances. Additionally, a function instance won't handle multiple requests at a time.
Given this, using beam is a bit of a waste in terms of individual instance scalability. That being said however, you might be able to use shared actor pools (e.g. caches, etc.) across all your functions. I want to emphasise the might as rapidly adding and removing nodes from the beam cluster might not work well or at all.
At the end of the day, the lambda model kind of supplants the actor model as your unit of messaging and concurrency, and so trying to mix the two isn't the best idea. If you want to use the beam on AWS I'd recommend sticking to ECS/Fargate/EKS. That being said, Elixir might be a nice match due to developer ergonomics, just don't expect to be able to drag and drop actor reliant features.
Could it not keep a naked beam and load your code once it knows that the request is for you? This it could keep a few spares around that are almost hot.
Maybe it shares a beam instance per AWS account. Definitely curious about the isolation level of lambda in general since the same comment also applies to the JVM.
[+] [-] bsaul|7 years ago|reply
Since the lambda model means concurrency is completely managed by the platform, why would you code in this language ?
[+] [-] mercer|7 years ago|reply
In particular the fault tolerance and process-based approach is just awesome. I regularly find myself writing code with just the 'happy path' (and only the obvious error handling), which just feels a ton nicer than the more defensive approach I generally take in other languages.
The tooling around the language is also really nice.
[+] [-] codeadict|7 years ago|reply
[+] [-] shshhdhs|7 years ago|reply
[+] [-] regularfry|7 years ago|reply
[+] [-] vazamb|7 years ago|reply
[+] [-] tirumaraiselvan|7 years ago|reply
[+] [-] skrebbel|7 years ago|reply
[+] [-] dalailambda|7 years ago|reply
Given this, using beam is a bit of a waste in terms of individual instance scalability. That being said however, you might be able to use shared actor pools (e.g. caches, etc.) across all your functions. I want to emphasise the might as rapidly adding and removing nodes from the beam cluster might not work well or at all.
At the end of the day, the lambda model kind of supplants the actor model as your unit of messaging and concurrency, and so trying to mix the two isn't the best idea. If you want to use the beam on AWS I'd recommend sticking to ECS/Fargate/EKS. That being said, Elixir might be a nice match due to developer ergonomics, just don't expect to be able to drag and drop actor reliant features.
[+] [-] nasmorn|7 years ago|reply
[+] [-] kenhwang|7 years ago|reply
[+] [-] thedoops|7 years ago|reply
[+] [-] cthalupa|7 years ago|reply
Nearly 100% of the lambda functions I've written could make use of erlang/elixir style pattern matching.