Re: Announcing kapture

[Evans Ye <ev...@apache.org>]

Great to hear you to bring up a new project to the community. Given the
goal of the project is for testing(which might be useful to test the
performance of distributed systems), the use of BPS, and the idea of
building cloud native big data stack, maybe it's a good thing we can add
this project as a sub-project of Bigtop. How do you and others think?

My considerations:
1. BPS code base has not been touched for a while, if we have a something
based on it, we can ensure it's working, and even bring back changes to it
to keep it stay maintained.
2. We can start from a separated testing sub-project but then driving the
path through an emerging point of cloud native big data stack.
3. K8S is for sure taking up the world. Would love to see something added
here in Bigtop. So that by time the community get's familiar with it.
4. Something new that we help leverage it for promotion and re-ignite the
community.

Evans

Jay Vyas <ja...@gmail.com> 於 2019年4月13日 週六 上午1:15寫道：

> Hi bigtop .... it’s been a while since peters news so figured I’d try to
> cheer everyone up with some news about my latest big data / kubernetes
> adventures.
>
> I recently got a new use case for some load testing in k8s networks, and
> wanted to have an intuitive way to measure overall cluster throughout on
> unhealthy clusters.
>
> So I thought I’d let folks know we’ve been using the BPS transaction queue
> for it .  The name of this new project I’ve been hacking on with a friend
> of mine, Ben Fenwick.
>
> https://github.com/carbonrelay/kapture
>
> It launches transactions into a Kafka spout and then reads them into
> various redis sinks.
>
> It’s a really cool way to bring down and load test the network, CPU, and
> memory on any cluster. For me, im using it to compare network performance
> in clouds vs bare metal.
>
> On this note , id still love to explore the idea of a next generation
> cloud native style big data stack , heavily focused on kv and messaging ,
> and I think a fork of this project could be the seeds of such a thing.
>