Well, I apologize for wasting a bit of your time. By putting httperf on one of the new machines and pointing it towards the other webserver, the results are a lot better. Since the tool was so small, it didn't occur to me how much of a little piggy it could be. Won't happen again, my quota of looking like a total idiot in a public forum is once per application.
I don't know if it will make a large change in the results, but I added the --hog flag since the number of connections is a fair bit larger. <new hardware, no load> httperf --server=test --port=80 --uri=/bg.gif --num-conns=31000 --hog Reply rate [replies/s]: min 2699.0 avg 2748.7 max 2798.4 stddev 70.3 (2 samples) Reply rate [replies/s]: min 2691.0 avg 2732.1 max 2773.2 stddev 58.1 (2 samples) Reply rate [replies/s]: min 2629.0 avg 2645.0 max 2661.0 stddev 22.6 (2 samples) Reply rate [replies/s]: min 2669.0 avg 2694.4 max 2719.8 stddev 35.9 (2 samples) httperf --server=test --port=3000 --uri=/images/bg.gif --num-conns=25000 --hog Reply rate [replies/s]: min 1317.3 avg 1323.4 max 1327.7 stddev 5.4 (3 samples) Reply rate [replies/s]: min 1324.5 avg 1326.7 max 1329.9 stddev 2.8 (3 samples) Reply rate [replies/s]: min 1323.5 avg 1333.5 max 1340.1 stddev 7.3 (4 samples) Reply rate [replies/s]: min 1321.5 avg 1334.8 max 1346.3 stddev 11.6 (4 samples) (hitting a blank ruby controller & blank page) httperf --server=test --port=3000 --uri=/test --num-conns=6000 --hog Reply rate [replies/s]: min 301.8 avg 320.9 max 336.2 stddev 17.5 (3 samples) Reply rate [replies/s]: min 318.0 avg 332.1 max 344.2 stddev 13.2 (3 samples) Reply rate [replies/s]: min 320.6 avg 321.7 max 322.8 stddev 1.1 (3 samples) Reply rate [replies/s]: min 328.4 avg 333.7 max 338.6 stddev 5.1 (3 samples) <existing hardware, possible light load, testing just apache> httperf --server=old --port=80 --uri=/bg.gif --num-conns=8000 --hog Reply rate [replies/s]: min 706.6 avg 707.0 max 707.6 stddev 0.9 (2 samples) Reply rate [replies/s]: min 663.2 avg 669.7 max 675.9 stddev 8.6 (2 samples) Reply rate [replies/s]: min 627.0 avg 660.5 max 693.2 stddev 47.1 (2 samples) <existing hardware, possible light load, same blank controller / code as above> httperf --server=old --port=3000 --uri=/test --num-conns=3000 --hog Reply rate [replies/s]: min 138.6 avg 157.0 max 162.6 stddev 8.4 (3 samples) Reply rate [replies/s]: min 138.2 avg 153.7 max 163.2 stddev 13.6 (3 samples) Reply rate [replies/s]: min 147.0 avg 156.9 max 162.2 stddev 8.6 (3 samples) Alexey, >> to get a ballpark of what the apache max performance ought to be >Several thousand hits/sec. is the 2.7k replies from the straight apache about what you would expect, or is it still on the low side? The performance dropped by about 1/2 when going through Mongrel which Zed said is very good, so assuming that Apache is working well, thats reassuring. Bob Brazeau -- Posted via http://www.ruby-forum.com/. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Deploying Rails" group. To post to this group, send email to rubyonrails-deployment@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/rubyonrails-deployment?hl=en -~----------~----~----~----~------~----~------~--~---
