Community Session: Chat with the Fluid compute team!

[Music] hey I think we are live yeah all right hello welcome to this week's versel Community live session I'm Amy your moderator for today's session and I'll be joined today by the flu compute team if you feel comfortable sharing with listening from please let us know in the

chat of course feel free to share any comments questions throughout the session before we begin I just want to remind everyone of the community conduct you can find that linked up in the site and down below in the floter um and let's just continue to keep this a positive and respectful environment for everyone and with that out of the way let's welcome Mariano Doug Craig and here to talk about fluid compute guys welcome hello hello

everyone all right where shall we start shall we do like a quick round of intros my name is Mariano product manager for compute here at barcel now D oh I'm muted classic um nice to meet you everyone um so I'm Doug I'm one of the engineers who worked on fluid and is responsible for our new fluid compute I passess on the button to

jav hello I am jav I'm a software engineer here at bcel working as a tech the compute team and I also worked on fluid compute so pass on to Craig I'm Craig another developer who also worked on on fluid compute uh for the last about two years we've been working on it yeah yeah constant Evolution so it just goes on forever and ever nice yeah I think like that really summarizes what fluid is it's been like a constant

Evolution coming from servers up to today so I think like that is giving us like the right cue for udak to start demo and yeah yeah talk about the yeah so I was going to start with a quick demo of what fluid compute is um and specifically like contrasting it with traditional servers and with this kind of traditional serverless model um so hopefully everyone can see my screen um but what we have here is like a little demo that kind of reflects how servers

work and I'll flip between the different kind of I don't know paradigms whatever you whatever you want to call them um later on so you can see I have a server running it's always on um and you can see that it has like a you know I'm AC crewing usage the whole time and that usage is what I'm paying for so whenever I have a server I'm paying for it being on um if I send requests to it you can see these funky little bars come in you can see the sort of amount of CPU time and idle time that each request is doing

um I will say this is kind of like somewhat reflective of what happens when um you know on the versel platform when people make requests a lot of our hosting is like next s apps um and when you send a request it's going to do some CPU intensive work at the start to like route the request pass the data Etc um that's typically followed by something like a database call or an API fetch something like that which is iio bound or idle time so it's not actively using the CPU it's just waiting on like external resources um and then finally

once that IO is finished we typically see some like short period of rendering time so these requests that we're sending here are kind of Representative of what we typically see on on verel um and I mean when I say on verell I mean like you know all the people who host their applications on vers so it's a very typical like pattern um so as a traditional server if I send a bunch of requests all at once um what happens is if I send too many my server becomes overloaded and you see this kind of slowdown effect here where I'm actually