Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why aren't we all serverless yet?

varoa.net

Post reply on HN
Print view View original

51–60 of 137 posts

Re: Why aren't we all serverless yet?

#51

Post by bdcravens »

Trasmatta wrote: post #18

For me, it's because Rails has continued to be an excellent solution in every application I've ever needed to build, whether it's a project with 1 user or 10 million, or with a dev team of 1 or 100. Every time I try to solve a problem with anything other than Rails, I run into endless issues and headaches that would have been already solved if I just. used. Rails.

Even when you have a problem set bigger than Rails, keeping everything in the Rails world and using something like Sidekiq to manage most of the backend complexity. For many cases, reasonable polling works as well as event-driven architecture, but if you absolutely have to do that, one-off lambdas that talk to Sidekiq or other parts of your Rails-stack work well enough.

Re: Why aren't we all serverless yet?

#52

Post by RickHull »

badlibrarian wrote: post #24

Because we use Nix recipes to deploy our Datalog-ish backend connectors that talk to Amazon Elastic Beanstalk via a bespoke database we wrote in Julia that's deployed on Snowflake Container Cloud. But there's a missing backslash somewhere and nobody can find it because even ChatGPT cannot decipher the error messages. Maybe it's an expired certificate but the guy who knew how that stuff works built a 12,000 line shell…

The problem with astronaut architecture is that nobody tells you about (or has a handle on) all the space junk.

Re: Why aren't we all serverless yet?

#53

Post by tomnicholas1 »

nilamo wrote: post #37
baobun wrote: post #10

This misses the main factor, I think: Vendor lock-in. There is no unification of APIs - every provider has their own bespoke abstractions typically requiring heavy integration into further vendor-specific services - moreso if you are to leverage USPs. Testing and reproducing locally is usually a pipe-dream (or take significantly more effort than the production deploy). Migrating to a different cloud usually requires…

There are a few platform abstractions. Quarkus, a Java framework, has Funqy, an extension that abstracts the differences between something like aws Lambda and Knative triggers, and feels quite easy to use. https://quarkus.io/guides/funqy

In python there is lithops, which provides nice Executor primitives that can run on a wide range of cloud services (AWS lambda, GCF etc.)

https://github.com/lithops-cloud/lithops

Re: Why aren't we all serverless yet?

#54

Post by Olreich »

This article misses the most important reason to not use Serverless: Cost. It's way more expensive to run serverless than it is to run any other format, even something like AWS Fargate is better than Lambda if you keep your Lambda running for 5% of the time.

The second one is even more important though: Time. How many of my systems are guaranteed to stop after 15 minutes or less? Web Servers wouldn't like that, anything stateful doesn't like that. Even compute-heavy tasks where you might like elastic scaling fall down if they take longer than 15 minutes sometimes, and then you've paid for those 15 minutes at a steep premium and have to restart the task.

Serverless only makes sense if I can do it for minor markup compared to owning a server and if I can set time limits that make sense. Neither of these are true in the current landscape.

Re: Why aren't we all serverless yet?

#55

Post by frereubu »

crzylune wrote: post #40

Serverless doesn’t mean no-server. It means someone else’s server. Their system. Their rules. Their way or the highway. No thank you.

I hate the term "serverless". It's a misnomer to the extent that it feels like it was designed to deliberately mislead. Even vague consultant-speak like "externally provisioned infrastructure" would feel more accurate.
Post reply on HN