Any plans on bringing apache arrow based segment columns?

Apache arrow seems pretty promising on in memory columnar storage.
Is anyone looking into using that to store/read segments in druid?

Will there be any real gain from doing this?

I don’t think anyone is focusing on this at present. It’d be interesting to see some benchmarks though.

how much serde overhead does druid have while processing a query?

are there any benchmarks on cpu times for serde? or any metrics being collected which can be quickly referred to.

I am new to druid and can experiment on this.

Hey Vivek,

There are a bunch of benchmarks in the “druid-benchmarks” module that might be useful.