you are all rookies. you are all aguing about CPU cycles (someone mentioned C). CPU cycles is not the goal in scalability. is an opitimised assembler app serving a fictitious figure of 1000 pages at once scalable? no, because how do you then serve 2000?
assume slow technology A cans server 100 pages at once. assume faster technology B servers 200 pages at once. A is given more overhead so that it can serve 50 pages simulaneously on seperate servers. A is run on 10 servers, thus serving 50 * 10 = 500 pages at once. A is more scalable than B.