Amos Jeffries schrieb: > Can you grab some ball-park figures as described > http://wiki.squid-cache.org/KnowledgeBase/Benchmarks ? Here we go: Users - can't tell since it's an accelerator setup RPS - 417 req/s Hit Ratio Request Hit Ratios: 5min: 93.2%, 60min: 94.6% Byte Hit Ratios: 5min: 91.4%, 60min: 91.9% Avg. CPU Load - 8.31% All in all I reached an improvement of 50% in the CPU load compared to our old setup. The improvement dramatically increases in our prime time. Here some numbers(from our Cacti graphs): RPS - ca. 990 req/s Avg. CPU Load - 16%(old setup: 52%) I thought I'd add the outgoing traffic(in MBit/s): Max 42,2 Avg 19,3 Once again I want to point out the content we deliver. A lot of extremely small files: Read Histogram: 1- 1: 1846 0% 2- 2: 1782 0% 3- 4: 2884 0% 5- 8: 5495 0% 9- 16: 9707 0% 17- 32: 23024 0% 33- 64: 44926 0% 65- 128: 90100 0% 129- 256: 7880423 5% 257- 512: 49731503 33% 513- 1024: 2247722 2% 1025- 2048: 28785893 19% 2049- 4096: 39794361 27% 4097- 8192: 9743983 7% 8193-16384: 7580467 5% 16385-32768: 2112179 1% Let me know if you need any more information. Cheers, Markus
