SERVER-62554 will add some info to the collStats output

Summary

The purpose of this project is to reduce the frequency and duration of stalls (write unavailability) for large sharded clusters, ensuring that customers can scale to larger sizes.

• We will only split chunks that need to be migrated
• We will automatically merge contiguous chunks on the same shard
• We will stop using individual chunks as a way to balance a sharded cluster and instead will migrate ranges (partial chunks) of data

Motivation

• Fewer stalls from removing auto-splitting (causes refreshes).
• Lower read and write latency spikes during balancing due to fewer chunks.

Cast of Characters

• Product Owner: garaudy.etienne
• Project Lead: tommaso.tocci
• Program Manager: connie.chen
• Drivers Contact:

Documentation

Product Description
Scope Document
Technical Design Document