3 ms·
You should read up on why this think time exists. It has nothing to do with "emulating slow clients" and everything to do with not claiming "I have a fast datab
by knz42 7y ago
You should read up on why this think time exists. It has nothing to do with "emulating slow clients" and everything to do with not claiming "I have a fast database" by running gazillion txn/sec on just 1MB of data in RAM.
TPC-C requires that you increase the amount of "live data" if you want to display/advertise more performance. That's the benchmark's rule.
If you want to benchmark something else, that's fine, but then
1) don't call it "TPC-C"
2) don't compare with databases that play by the rules.
- dkhenry 7y agoI know why it exists, and my point is they are not trying to show a gazillion txn/sec on 1MB of data. You are looking at a dataset that is several TB's. They have far surpassed the point where a vendor is trying to cheat by putting all the data in L1 cache and claiming to be fast.