We have been exploring the tradeoff space between processes and functions, containers and lambdas, for a long time. We understand the problem extremely well, and we have seen and learned first hand what scales and what doesn't.

@rauchg

Are you planning to support deployment of, e.g., web services with multiple endpoints, implemented e.g. with Python Flask, without making changes to the code?

@raucgh

How about websocket over lambda ? I would be happy to refactor a monothlic to lambda, as soon as it supports websocket connection. Thanks.

@sisp

we are investing quite a bit into our Python builder right now, as a matter of fact. We have a lot of good news coming for the python community :)

@revskill

That's one of our top priorities to address. As the creator of socket.io, I know exactly what works and what doesn't. I have been researching realtime apps for a very long time now, and I'm confident Now is going to be the best place to host them and scale them to billions of users.

Here's the kind of interesting thing. Now is a unique company in that we're never going to lower the bar of ambition. We think we can give you the best of both worlds: ease of use and planet-wide scalability.

Throughout a journey like this, you'll find that some initial assumptions sometimes need to be tweaked.

In this case they're being tweaked so that you never worry about your next app taking over the world and dealing with billions of requests per day.

Please

@rauchg

don't forget the Apollo community :D I know you guys have a lot to handle right now, but we were super excited to host everything on now. I hope with get more details on this new approach on the coming weeks

It's really not a bad tradeoff to make, all things considered.

@zouhair

I meant it when I wrote in the blog post that my main priority is to address all the customer workloads perfectly.

I also wouldn't be calling this 2.0 if we had to reject entire classes of interesting applications.

Also, everyone is free and welcome to email me directly at [email protected]

And I'm happy to jump on a call with any teams that need further reassurance.

I am sorry. I must be missing something. What does this mean exactly? "00% confident that this is the right direction for the platform"? Does this mean you do not believe that docker is a good tool for production deployments? The only alternative interpretation I can make, is that the Zeit Now platform has many more non-docker than docker deployments, and then, it makes sense from Now's perspective, but it also means a "hard goodbye" to anyone seeking docker builds.

@rauchg

what is your approach to "address all the customer workloads perfectly"? Is there a place we can submit use-cases that don't have an apparent fit for 2.0?

@Matt

we are aware of several of them, please let me know if I forgot any:

- GraphQL servers like Apollo Server

- WebSocket subscriptions

- Control loops (e.g.: tailing a cursor from a database)

- Cron jobs

All of these will be addressed, and others as well!

@rauchg

Can we still create an entire application on now 1.0 and choose a 1.0 plan? I definitely can't use 2.0 in it's current state.

@rauchg

I think , the now.json should accept all versions. It will make all of us happy !

What about elasticsearch?

@revskill

it does

@akryum

that was never possible to host to begin with, we only ever supported non-durable workloads

But we definitely want to make it super easy to connect and set up elasticsearch clusters