We have installed few linux servers , all having same hardware and same OS. The hardware is DL380 Proliant. We have RHEL with the following details
Kernel 2.6.9-34.0.2.ELsmp #1 SMP Fri Jun 30 10:32:04 EDT 2006 x86_64 x86_64 x86_64 GNU/Linux

It has 16GB RAM and 2 Dual core CPUs. We have the below problem with 2 servers. One server has Oracle and other server has a java process.

The server starts perfectly and after some period(some times 1 day, sometimes 3 days, sometimes 3hrs of server startup) of time, it just hangs. Hangs means that Im not able connect(telnet, sqlplus or anytype connections to the server). The exsiting open oracle connections does not return any data.

It would be nice if someone could tell me what areas I should be looking for and how to debug this issue?

NOte: I did some analysis and found that the memory is pegging at 99.5% most of the times and approx 10% of SWAP is used.

It's normal for Linux to use most of the existing RAM for cache/buffers but 10% swap is a little bit on the high side. Have you run vmstat or iostat to see how much swapping is regularly going on?

That being said, a high swap load shouldn't cause a systewm crash unless you completely exhaust swap space. Have you noticed any commonality in the hangs? Same time of day? Heavy app/db traffic?

Can you hook up a keyboard/mouse to the machine and see if there's anything on screen (i.e. kernel panic or something similar)?

Try to record 'top' output. Is there any sharp rise in memory usage before hang? Looks like system suddnly starts to swap and just lacks machine time to do anything useful (swapping and waiting for swapped paged to load are probably non-interruptible and so effectively hang a box, while it is formally responsive (just really very slow).