When you say 'goes down' what exactly do you mean? I am guessing if you are replacing hardware it is not just one service (e.g. Apache) going down but the whole box. Is there anything in the logs (/var/log/messages or dmesg) to indicate a hardware problem? Anything on the console (remote or local)? This is where we would start looking for clues to confirm that it is in fact a hw problem. You might want to keep two windows open on it to see what is going on. An ssh window running top and a remote or local console window for console messages. Then when it goes down you might have an idea why. No messages in the logs or console, try replacing the RAM again (with good RAM e.g. Micron/Crucial). There is a lot of bad RAM floating through the marketplace and the symptoms are usually complete crash with no error messages. Unlike a hard drive problem, when the RAM crashes the machine there frequently will be nothing in the logs or on the console. Good luck.