The current workload for testing WT performance with eMRC=true described in WT-6661 and WT-6776, test a couple of different workload mixes:

• 100% insert
• 50/50 mix of find and update
• 90/10 mix of find and update
• 100% delete

All of these phases operate on a single collection.

We might discover different issues with a different mix of benchmark operations. In particular we might want:

• 100% update.  This seems to be the most stressful thing we can do to the history store (see HELP-18323).
• 100% reads.  Contrary to the previous item, this should be the least impacted by the history store.
• Include some insert and delete operations in the mixed workloads
• Spread the work across multiple collections, either uniformly or not.
• Vary document sizes. The current test uses documents between 50 and 200 bytes.  We could look at larger documents or various mixes of larger and smaller documents.

This ticket is not intended to implement and test all of the above.  The goal is to identify a small number of useful ways to extend the benchmark, implement them, and run the corresponding tests.  The list above is simply meant to enumerate some possible directions.