Dynamic colocate in weaver-kube or weaver-gke

[Suresh Kumar R]

I tried looking for code in github to figure out if there is controller code to dynamically orchestrate colocation of components based on metrics, latency and dependency of components along with scaling.

However, it seems that there is no auto scaler written for colocation based on metrics and that colocation is given in spec as in link https://github.com/ServiceWeaver/weaver-kube/blob/main/internal/tool/deploy.go 

Configure groups of collocated components. E.g., if you want two components
       "Foo" and "Bar" to run in the same OS process, you can put them in the
       same group. E.g.,
       groups:
         - name: group1
           components:
             - Foo
             - Bar

So, is there any plans to make dynamic scaling controller that works with HPA or KEDA to decide colocation of components dynamically?

or Am I missing some codebase that exists to colocate components dynamically based on metrics?

[Robert Grandl]

Hi Suresh,

You are right. Currently, the colocation groups have to be manually specified in the config. 

Dynamic colocation based on load, dependency graph, etc. is something we've been considering. However, as we haven't received any requests to support this scenario yet, it hasn't been a short-term priority for us.

Based on the feedback we've received so far, users want control over how they collocate components; otherwise, it feels too magical.

Thanks,

- Robert

--
You received this message because you are subscribed to the Google Groups "Service Weaver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to serviceweave...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/serviceweaver/0aaa162e-4723-45d8-bfc3-b1e213e38ba6n%40googlegroups.com.