The Spike is ready and had involvement from Ryan's team already. Creating this tickets for bookkeeping.

Description of Linked Ticket

Summary

Prototype and evaluate approach(es) for cluster to cluster replication to handle concurrent document movement between shards on the source cluster. This will facilitate follow-on work to support 1) concurrent chunk migrations/balancer and 2) concurrent changes in document shard key values on the source cluster for Cluster to Cluster Replication.

Motivation

As in the proposed architecture for Cluster to Cluster Replication, when the source cluster is a sharded cluster, we would want to establish per-shard change streams via the source mongos(es). And if we want to increase parallelism and apply change events from different shards concurrently, we may get out-of-order (conflict) change events if a document is moved between shards (e.g. by a chunk migration). This project is to prototype and evaluate solutions to support sharded clusters for Cluster to Cluster Replication with the balancer enabled. And this work will allow us to compare the complexity, feasibility and cost effectiveness of different solutions. And this project will also serve the purpose of coming up with a concrete design for the project Replicator: Support Conflict Resolution during Sharded Cluster Replication (REP-33) and identifying server requirements/dependencies.

Cast of Characters

• Product Owner: Alan Zheng
• Project Lead:
• Program Manager:
• Drivers Contact:

Documentation

Scope Document
Technical Design Document
Product Description