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

91–100 of 137 posts

Re: Why aren't we all serverless yet?

#91

Post by 9rx »

qaq wrote: post #88
9rx wrote: post #84

Earlier quoted context omitted.

> Serverless is all about outsourcing the infrastructure for scaling a micro service. Technically, it is all about removing the server from your application. The name literally tells you so. It is true that removing the server can offer some benefits in the scaling realm. In particular, it allows you to scale to 0 now that you no longer have to keep the process alive to serve requests. Of course, that is not tradeoff…

Scaling to 0 is pretty much the only benefit you get. Which is not much of a benefit at any reasonable scale. Fargate is same firecracker VMs as Lambda they just don't scale to 0.

And even then it is not strictly necessary. Cloud Run comes to mind as offering the ability to scale to zero, yet allows (maybe even requires?) maintaining the server in the application. The real benefit of "serverless" is that you no longer have to worry about the server, it being removed. Granted, that is not a big worry much these days with all the great server frameworks available.

Re: Why aren't we all serverless yet?

#92

Post by woleium »

Vt71fcAqt7 wrote: post #73
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…

Google cloud run and Azure container apps both let you run an arbitrary docker image without having to deal with custom setups. Both scale automatically so they are serverless. AWS has apprunner but it doesn't scale to zero.[0] [0] https://github.com/aws/apprunner-roadmap/issues/9 (amusingly the issue OP posts on HN)

There is also knative, which cloud run is based on

Re: Why aren't we all serverless yet?

#94

Post by freefaler »

tw04 wrote: post #47

Because it's more expensive for less performance with less control. If it were 5% worse performance for 7% more cost, most people would probably not bat an eye. When it can be 50% less performant for 200% more cost, eventually someone is going to say: sure there's overhead to owning that but I will be at a major competitive advantage if I can do it even just OK. And it turns out for most businesses doing it at the sc…

Indeed... I've run on Hetzner for 20 years with triple redundancy and VPS for batch processing/CI and some internal tasks. My costs are fixed and only on very big database alters/upgrades/migrations our service has any downtime.

I have a friend who recently made a stupid bug in his processing pipeline on AWS. He woke up on morning and saw a message from his bank that his CC was over the limit.

When we have a bug, our Nagios send us a message that responces are more than 150% of average and we do a rollback.

So it's not only the risk of vendor lock-in, but also in surprising bills and policy changes, updates and other 3-rd party risks you end up with.

Re: Why aren't we all serverless yet?

#95

Post by hylaride »

tw04 wrote: post #47

Because it's more expensive for less performance with less control. If it were 5% worse performance for 7% more cost, most people would probably not bat an eye. When it can be 50% less performant for 200% more cost, eventually someone is going to say: sure there's overhead to owning that but I will be at a major competitive advantage if I can do it even just OK. And it turns out for most businesses doing it at the sc…

This highly depends on workload. We migrated a service that generates terrabytes of content to send to customers each day. We moved the content generation from J2EE to java lambas and our costs went from $6K/month (on savings plans, evemn) of ec2 to ~$400/month in lambda, sqs, and elasticache/redis costs and the work was done in 1/8th the time. Mind you, our content is highly bursty where we need to be able to generate the content within seconds seconds of initiation.

Serverless also means a lot of things. We also serve static content from an S3 bucket and cloudfront. Nothing else to manage once its setup.

The flip side of serverless is you really do need to think of state yourself. The J2EE code was rock solid in reliability, including recovering from almost every kind of issue you can imagine over a decade (database, connectivity, software crashes).

Re: Why aren't we all serverless yet?

#96

Post by qaq »

9rx wrote: post #91
qaq wrote: post #88

Earlier quoted context omitted.

Scaling to 0 is pretty much the only benefit you get. Which is not much of a benefit at any reasonable scale. Fargate is same firecracker VMs as Lambda they just don't scale to 0.

And even then it is not strictly necessary. Cloud Run comes to mind as offering the ability to scale to zero, yet allows (maybe even requires?) maintaining the server in the application. The real benefit of "serverless" is that you no longer have to worry about the server, it being removed. Granted, that is not a big worry much these days with all the great server frameworks available.

I’d much rather write 20 lines of boilerplate code and a Dockerfile than deal with 100s lines of CDK code, distributed tracing, and the associated observability challenges.

Re: Why aren't we all serverless yet?

#97

Post by freefaler »

kasey_junk wrote: post #56
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.

How much do your data centers cost to build roughly? How do you get global bandwidth with out peering?

How much would a VPS or a rented server cost where you can boot your own OS and be the sole tenant of the SSD and don't fight with the IOPS of the other videoconverting dude using the same machine?

Re: Why aren't we all serverless yet?

#98

Post by sitkack »

qaq wrote: post #35
sitkack wrote: post #6

Earlier quoted context omitted.

Without a link and a breakdown that makes no sense. We switched from blue to square and saw a honey suckle savings. Amazon runs both and serverless is a billing model. Many serverless runtimes consume containers. Serverless, like microservices are a design philosophy.

the pricing is different for the same amount of compute is that news for you? Lambda is priced an order of magnitude higher than fargate which is priced significantly higher than EC2. For small scale workloads your TCO might be lower with higher level abstraction.

It is a billing switch, crazy that Amazon would publish a report of them do a rewrite of an application when the distinction is one of book keeping.

You would think they would want to sell their expensive solution.

Re: Why aren't we all serverless yet?

#99

Post by isoprophlex »

recursivedoubts wrote: post #2

they constantly try to escape from the complexity outside and within by dreaming of abstractions so perfect that no one will need to be good but the latency that is will shadow the "simple" that pretends to be

grug wonder why big brain take hardest problem, factoring system correctly, and introduce network call too

seem very confusing to grug

Re: Why aren't we all serverless yet?

#100

Post by Brian_K_White »

9rx wrote: post #75
frereubu wrote: post #55

Earlier quoted context omitted.

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.

It seems it is only a misnomer if you are too young to remember how these types of applications used to be written. They weren't always servers. In the early days they were subprocess modules[1]. "Serverless" is a return to the subprocess model, seeing the application lose the server, or to put it another way the application is less a server. This must be why they say programming is dead once you turn 40: You can no…

I'm from even before that, and it makes no sense to me.

I understand both what you say and what "serverless" commonly means, I'm just saying it's essentially arbitrary. A symbol with no etymology that holds water.

Post reply on HN