When you download and install the JetStress tool, one of the accompanying files is Microsoft Exchange Server JetStress.chm which is, of course, the associated help file for the tool.
Within the documentation, under the section titled "Testing for CCR, LCR, and SCC Configurations" there is the following text:
"Run the same Jetstress configuration (for SG/DB count, database sizes, and threads) against both nodes of the CCR cluster. The number of Jetstress threads should be increased to simulate the 3x I/O related to the CCR log replay overhead. For adequate performance, each cluster node must be able to achieve 3x the I/O of the expected user profile within the expected latency (20 ms)."
This refers to the Exchange 2007 RTM issue whereby the passive node of a CCR environment required 2-3x the I/O of the active node. This issue was addressed in Exchange 2007 SP1 and so the above text is not relevant for SP1.
Hopefully this text will be changed during the next documentation update.