Drill hasn't been quite as performant as I was hoping (yet, still trying).
I'm curious if the drill-bit sizing has anything to do with this. Our admins mange the cluster-side settings though so I'm going to have to rely on asking questions over experimenting.
- Our queries are finishing, they're just a little slow. So, they are not outright failing.
- Our Drill UI shows Maximum Direct Memory 8,589,934,592.
- What is the default size of a drill-bit processes?
- What is "Maximum Direct Memory" exactly?
- Assuming our queries are finishing, will giving drill more memory possibly improve their performance? If so, how do you size the memory?
- Can I see if drill is expending its memory in any easy way (e.g. flushing to disk or whatever it does)?