Hi! How do u think, it will be good idea to use KT...

# ktor

[qwert_ukg]

Hi!
How do u think, it will be good idea to use KTOR as entry point of microservice architecture?

[orangy]

That’s one of the use-cases we are targeting, especially with Ktor’s own HttpClient. It should provide full async pipeline across the microservices.

[qwert_ukg]

hope it will be released soon

[orangy]

Do you really need a version that reads 1.0?

[qwert_ukg]

not me, my superiors. I know what u already use it in JB, but for them 1.0 more major argument then it

[dave08]

I'm just looking forward to async db clients and it'll be my choice for our microservices over nodejs and vert.x...! Also vert.x has an eventbus, does the httpclient do the same (use a socket for communication with serializable objects for requests and responses)?

[orangy]

@qwert_ukg That’s weird, but perfectly understandable, unfortunately… Code doesn’t become more stable, more safe or anything if we change it from 0.9 to 1.0.

[orangy]

@dave08 HttpClient is an HTTP client, you know 🙂 It’s not an event bus, it’s not cluster streams, etc. With HTTP you can use web sockets to establish connection and exchange data, if you wish, but using some specifically designed system like Kafka would be better.

[dave08]

You're probably right that there's alternatives, but I was wondering about something simple integrates into Ktor.

I find that vert.x is too opinionated and overkill when using external orchestration like Docker Swarm or Kubernetes... and it's such a common use case for microservices that Ktor might have some easy and reliable way of communication between microservices based on it.. it doesn't need clustering features since orchestration takes care of that. I guess it's not critical, but could be nice.

I currently have all our microservices in vert.x with a wrapper layer, waiting for when Ktor will have at least the async db clients so I can switch...

[orangy]

Well, we don’t have capacity to implement all of these… Performant and reliable event bus is not an easy task, and we believe existing solutions are pretty good. It’s not our goal to reimplement all the existing tools, it’s the other way around, we want more or less anything to be used with ktor. A little bit of glue might be welcomed, so if you want to make Kafka Feature, you’re welcome 🙂

[dave08]

Now that I think of it, I see how involved it would be🙂, but making a Kafka Feature is an interesting thought..

When I switch over with the advent of async dbs, I might think of it...