{"items": [{"author": "Jan-Willem", "source_link": "https://plus.google.com/100580955183019057735", "anchor": "gp-1480777306386", "service": "gp", "text": "Depends on exactly what you're trying to test, of course.  For normal operation but with high load, replaying a trace of actual traffic at high speed works well for exercising caching.\n<br>\n<br>\nIf you're trying to load the parts of the system protected by the cache, though, you actually \nwant\n to do something like random url access.", "timestamp": 1480777306}, {"author": "Jeff&nbsp;Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1480785515884", "service": "gp", "text": "@Jan-Willem\n\u00a0Replaying traces is good, and is probably better than using \"here's what my traffic looked like over the last two years\" because it handles changes in what's popular over time.  It does require more data to run your test, though, and is hard to handle for new systems.\n<br>\n<br>\nAnd yes, if you're trying to test what's under the cache you should do something to turn the cache off or render it ineffective.", "timestamp": 1480785515}]}