neon VS glommio

If you're reading this you probably got a really steep bill from Neon after finding yourself on their "Scale" plan. If you do want to stay with Neon but avoid surprise bills then go to the Plans page and choose what you actually want.

Such is the case with Neon, a serverless Postgres service, that went generally available on April 15. Congrats Nikita Shamgunov and team on the launch. When I saw the announcement, I knew I had to try it out for myself and report back with my findings.

I want to use this as a chance to bring attention to a GitHub issue that I think would help reduce friction for Neon:

https://github.com/neondatabase/neon/issues/4989

If the Neon driver were to allow us to easily pass in a localhost connection, the development and test experience would be easier. Perhaps Neon could swap to something like this internally: https://github.com/porsager/postgres.

Having run a local dev environment connected to Neon and tests connected to Neon got in our way of adoption. We'd prefer to develop and run tests against a regular Postgres localhost database.

To the PMs of Neon, put yourself in the shoes of a new developer thinking of giving Neon a try. What changes will I have to make to my code and my development workflow?

Neon is an open source and cloud-native serverless database platform that focuses on simplicity and ease of use. It supports Postgres databases and offers built-in features like bottomless storage, autoscaling, and branching.

For MySQL, we've got PlanetScale, and for PostgreSQL, there's Neon.

8. https://neon.tech/As you might know not one tool fits all, I still have strong preferences for the following. It helps me get going faster and get things done right first time and helps in ease of maintenance.

Language: Typescript.

Migrated away when they removed the free tier, ended up using https://neon.tech/

Cool! At Neon[0], I work full time on our custom postgres proxy[1]. It's a very nice protocol to work with, although our usecase is quite a bit more complex compared to the ideas presented in the post.

Neon databases scale to zero, so the proxy needs to spin up databases on the fly. The proxy doesn't do that but it knows if the databases is running and asks our control plane to schedule it if it isn't. It's a fun service to maintain.

The biggest pain is error handling. Postgres is really bad for error messages and codes. The only available code we can use is usually protocol violation...

[0]: https://neon.tech/

Neon - PostgreSQL

I used glommio as the async executor (instead of something like tokio), and it is wonderful. For people wondering whether it's "good enough" or to use C++ and seastar (as I have thought about a lot before starting this project), take the leap of faith, it's fast - both in terms of run time and to code.

My understanding is you always need a runtime, somethings needs to drive the async flow. But there are others on the market, just not without the.. market domination... of tokio.

https://github.com/smol-rs/smol looks promising simply for being minimal

https://github.com/bytedance/monoio looks potentially easier to work with than tokio

https://github.com/DataDog/glommio is built around linux io_uring and seems somewhat promising for performance reasons.

I haven't played with any of these yet, because Tokio is unfortunately the path of least resistance. And a bit viral in how it's infected tings.

I think you missed one which is based on io_uring [1].

In my benchmarks with a slightly tweaked version it was 2x faster than Nginx and and 30x faster than Python's SimpleHttpServer.

[1] https://github.com/DataDog/glommio/blob/master/examples/hype...

b) It's proven now e.g Seastar, Glommio that the fastest way to run a multi-threaded application is to have one instance with one thread pinned per CPU core. Then to have fibers/lightweight threads on top handling all of the asynchronous code. Your approach of lots of instances is the slowest so there will be a ton of unnecessary thread context-switching.

I assume Tokio itself, see e.g monoio or glommio, but also Seastar for C++.

https://github.com/DataDog/glommio Rust thread per core library.

> Few of us have really figured out io_uring. But that doesn't mean it is slower.

seastar.io is a high level framework that I believe has "figured out" io_uring, with additional caveats the framework imposes (which is honestly freeing).

Additionally the rust equivalent: https://github.com/DataDog/glommio

This use case is perfect for https://github.com/DataDog/glommio which is a thread-per-core runtime that is appropriate for latency sensitive code.

It's worth mentioning: Under "Async Executors", for "io_uring" there is only "Glommio"

I recently found out that ByteDance has a competitor library which supposedly has better performance:

https://github.com/bytedance/monoio

https://github.com/DataDog/glommio/issues/554

FYI, Datadog has a Rust library for scheduling things to run thread-per-core with io_uring

It'd be really useful for DB use cases:

https://github.com/DataDog/glommio