Serverless Servers Exist (this isnt a joke)

Рет қаралды 8,599

Күн бұрын

Пікірлер: 48

@MelkeyDev 3 ай бұрын

I hope you guys enjoy this video! Shoutout to the serverless servers that are not servers you own - but servers THEY own!

@SR-ti6jj 3 ай бұрын

Tailwind > Postegres

@vaibhav5783 3 ай бұрын

HTML > Rust

@michalzarddev 3 ай бұрын

@@vaibhav5783 CSS > SQL

@IndigoTeddy 3 ай бұрын

NeoVIM > MongoDB :Kappa:

@RomanAvdeevX 3 ай бұрын

@@vaibhav5783 Rust > Adobe

@peterszarvas94 3 ай бұрын

farming > javascript

@Kane0123 3 ай бұрын

A fleet of MacBooks in a closet - if it ain’t racked it ain’t a server

@Kairulol 3 ай бұрын

>be me >serverless server supervisor >in charge of making sure the serverless server is, in fact, serverless

@guillemgarcia3630 3 ай бұрын

2024 is the year of the serverlesslessness

@beachfeet6055 3 ай бұрын

Threaded Functions or as mentioned earlier, Scaled Functions. Being pedantic, I would avoid Concurrent Functions as concurrency is not parallelism, which seems to being implied here, simultaneous threads of execution in a function.

@emil_l889 3 ай бұрын

Mfs will do anything to use nodejs

@ooijaz6063 3 ай бұрын

Look how much effort they put into having type safe APIs with auto generated clients. Can't wait until they discover OpenAPI.

@dandandev 3 ай бұрын

@@ooijaz6063tRPC?

@masterflitzer 3 ай бұрын

@@ooijaz6063 openapi tooling still ain't as good as trpc, as much as i love openapi spec

@flarebear5346 3 ай бұрын

Next hear they will find out you can have 0 startup delay by keeping the process running on a server

@masterflitzer 3 ай бұрын

lmao

@nitinsanghi4652 3 ай бұрын

love the comment bottom of the video: "chatgpt, generate seo keywords and jumble them with technical words". awesome!

@flamakespark 3 ай бұрын

Vercel just invented a hypervisor for Lambdas. Damn, that's so genius 👏👏👏

@festusyuma1901 3 ай бұрын

It's a good feature, yes. but if you feel like you need this particular feature, I think it means it's time to use an actual server, lol I honestly don't see the benefit of locking yourself into vercel this much when you could do this on an actual server, since the audience they're targetting are people that actually need the concurrency

@C4CH3S 3 ай бұрын

Vercel is a platform that could be useful for maybe a handful of people, but somehow every sucker out there is using it.

@peterszarvas94 3 ай бұрын

cant wait to serverless serverless

@mshubitidze 3 ай бұрын

ion get it, why are separate lambda sandboxes waiting for the first one to finish?

@IndigoTeddy 3 ай бұрын

@@mshubitidze they aren't. If I understand correctly, Melkey was trying to explain that you had to spend the cost of running 3 separate lambdas for one request each (same costs running the same lambda three times due to the number of combined seconds from all instances), while this new paradigm allows paralellizing tasks on the same instance, meaning multiple tasks will only take up the concurrent time spent using the instance instead of the sum of times per request.

@mshubitidze 3 ай бұрын

@@IndigoTeddy@5:44 he explicitly says time to process request and calculates 100ms * 3, that’s what i didn’t get

@mshubitidze 3 ай бұрын

@@IndigoTeddy ooo, i get it, 300ms of compute time, measure that lambda is charged by, thanks

@AdamFiregate 3 ай бұрын

So basically scalable functions.

@erikslorenz 3 ай бұрын

I've just never been down with lambdas

@wk-kln 3 ай бұрын

Clientful Client

@AshesWake-sf7uw 3 ай бұрын

it should not be serverless, rather LESSSERVER 😂 get it? get it?

@mattymattffs 3 ай бұрын

Or you could just use real backend languages

@hebozhe 3 ай бұрын

Jaishless jaish.

@aliasjon8320 2 ай бұрын

The year of serverlesslessness is nigh

@Dom-zy1qy 3 ай бұрын

I don't know what this is supposed to be, but it sounds very VERY expensive. Edit: I read the blog post and it's actually just an improvement to their service, and makes it cheaper unless you're doing CPU bound workloads, in which case it would add additional latency, but at that point why are you even deploying to vercel? Though, from my understanding, this probably won't be beneficial to smaller projects. You won't benefit from concurrency if never need to handle more than 1 request concurrently.