Before we can build a performance test we need to know how to update the configuration property, how to monitor when it has taken affect and how to measure the time taken to replicate partition data.

Outcome of this spike task should be to have written down:

• Steps for how to update num.replica.fetchers in RHOSAK
• Steps to determine the current value of num.replica.fetchers in RHOSAK so we know when it has changed
• Steps to measure the time taken by Kafka to replicate partition data from the leader to one of the follower brokers

For the third outcome some potential options to investigate:

• Broker or client level metrics that give some information
• Running a producer and seeing how long it takes for the acknowledgement to come back
• Running a producer and consumer and seeing how long it takes for the consumer to see the record (this wouldn't be that accurate but would indicate a rough scale of how much changing the config impacts the time)