Improve Dataset Service Performance and Stability
Description
Release Notes
None
80% Done
is blocked by
relates to
Activity
Show:

Sreevatsan Raman November 4, 2016 at 10:45 PM
We have addressed blocker issues Server-side caching and added benchmark. Marking it as Major.
DatasetService has been the cause for various issues. Main areas of improvement needed:
Performance: It can currently not handle many requests, when many flowlets or service handlers start at the same time, some of them get timeouts for their RPC calls.
Stability: More graceful handling of transaction failures, high availability
Operability: Improve error handling, return meaningful errors, insights into error rate, perhaps metrics, more meaningful logs