Back in '99, Philip Greenspun wrote: >America Online is fielding 28,000 hits per second across all of its >various Web services and servers
I am curious if anyone knows what the configuration looked like at that time. How many servers was AOL using? What were they? On a somewhat related topic, I would like to know how many persistent connections a cough, typical linux server, cough, might be expected to reliably service, and what are the limitations involved. I've visted the C10K pages and the like, and ran across a quote attributed to Alan Cox suggesting that each connection requires 20K in the kernel (and a rejoinder of well that was dumb then) suggesting that 1000 connections would require 20M just for TCP's use. And it appears from a few years ago that a few hundred was okay, but more than a thousand was considered dicey. But that was a few years and a few kernels ago. What's the scoop now? How many persistent connection might a uniprocessor Xeon 1.6Ghz PIII with lots of memory reliably handle? Semi-informed guesses are what I am seeking. Is it 1K, 2K, 4K, 8K, 16K, 32K, ...? Thanks, Jerry
