That is a rather big topic. I use test strategies rather than test plans, but after several long threads worth of discussion, that seems to be purly a semantical distinction.
At the highest level, I have 5 sections in my Perf Strategy docs.
1) Intro, scope, etc.
2) Performance Requirements/Acceptance criteria
3) User/system models of activity to be tested
4) Verbal descriptions of the scripts to be developed to execute models, to include:
- Metrics to be collected
- Names of timers for each page
- user delays between pages
- all data requirements
- other relevant stuff
- Scheduled tests
- How bottlenecks will be handled
Please ask more questions if you want more detail.