If we're talking single server then I agree, but that wasn't what I meant. It looked to me like he was testing a load pattern that would correspond to a single *client*. And that surely isn't a good way to do scalability testing, unless your approach to scaling is one machine per user. (Which works pretty well for some applications, e.g. Office... But it's not what most people think of when the term 'scalability' is bandied about.)
-- Ian Griffiths DevelopMentor -----Original Message----- From: Thomas Tomiczek [mailto:[EMAIL PROTECTED] > -----Original Message----- > From: Griffiths, Ian > > But then you say that you're interested in all processes > being on the same machine, which implies that you don't in > fact care about scaling beyond, well, one. So what exactly Carefull here, Ian :-) Could be part of a larger architecture. I was designing something a,long these liens one (search engine like). We had independent computers which lived with copies of the data. Scalability was defined in two dimensions here, which were orthogonal: * How much can one computer handle. * Put more computers online, working independent. Priogrammers were ONLY optimising point 1, as poin t 2 was simple. This is applicable for a LOT of things: * News sites * Online shops (catalog part) * Search engines Etc. Don't assume he does not care about scalability JUST becase he optimises one computer scenarios. Thomas Tomiczek THONA Software & Consulting Ltd. (Microsoft MVP C#/.NET)
