Low Performance for big hospital server ..

I try to adjust my server for a couple of weeks with some sucess but it still
slow when the server has stress in the moring from many connection . I used
postgresql 7.3.2-1 with RH 9 on a mechine of 2 Xeon 3.0 Ghz and ram of 4 Gb.
Since 1 1/2 yr. when I started to use the database server after optimizing the
postgresql.conf everything went fine until a couple of weeks ago , my database
grew up to 3.5 Gb and there were more than 160 concurent connections.
The server seemed to be slower in the rush hour peroid than before . There
is some swap process too. My top and meminfo are shown here below:

207 processes: 203 sleeping, 4 running, 0 zombie, 0 stopped
CPU0 states: 15.0% user 12.1% system 0.0% nice 0.0% iowait 72.2% idle
CPU1 states: 11.0% user 11.1% system 0.0% nice 0.0% iowait 77.2% idle
CPU2 states: 22.3% user 27.3% system 0.0% nice 0.0% iowait 49.3% idle
CPU3 states: 15.4% user 13.0% system 0.0% nice 0.0% iowait 70.4% idle
Mem: 4124720k av, 4085724k used, 38996k free, 0k shrd, 59012k buff
3141420k actv, 48684k in_d, 76596k in_c
Swap: 20370412k av, 46556k used, 20323856k free 3493136k
cached

PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND
16708 postgres 15 0 264M 264M 261M S 14.7 6.5 0:18 2 postmaster
16685 postgres 15 0 264M 264M 261M S 14.5 6.5 1:22 0 postmaster
16690 postgres 15 0 264M 264M 261M S 13.7 6.5 1:35 3 postmaster
16692 postgres 15 0 264M 264M 261M S 13.3 6.5 0:49 1 postmaster
16323 postgres 16 0 264M 264M 261M R 11.1 6.5 1:48 2 postmaster
16555 postgres 15 0 264M 264M 261M S 9.7 6.5 1:52 3 postmaster
16669 postgres 15 0 264M 264M 261M S 8.7 6.5 1:58 3 postmaster
16735 postgres 15 0 264M 264M 261M S 7.7 6.5 0:15 0 postmaster
16774 postgres 16 0 256M 256M 254M R 7.5 6.3 0:09 0 postmaster
16247 postgres 15 0 263M 263M 261M S 7.1 6.5 0:46 0 postmaster
16696 postgres 15 0 263M 263M 261M S 6.7 6.5 0:24 1 postmaster
16682 postgres 15 0 264M 264M 261M S 4.3 6.5 1:19 3 postmaster
16726 postgres 15 0 263M 263M 261M S 1.5 6.5 0:21 3 postmaster
14 root 15 0 0 0 0 RW 1.3 0.0 126:42 1 kscand/HighMem
16766 postgres 15 0 134M 134M 132M S 1.1 3.3 0:01 2 postmaster
16772 postgres 15 0 258M 258M 256M S 1.1 6.4 0:04 1 postmaster
16835 root 15 0 1252 1252 856 R 0.9 0.0 0:00 3 top
2624 root 24 0 13920 7396 1572 S 0.5 0.1 6:25 1 java
16771 postgres 15 0 263M 263M 261M S 0.5 6.5 0:06 0 postmaster
26 root 15 0 0 0 0 SW 0.3 0.0 3:24 1 kjournald
2114 root 15 0 276 268 216 S 0.1 0.0 2:48 2 irqbalance
1 root 15 0 108 76 56 S 0.0 0.0 0:07 3 init
2 root RT 0 0 0 0 SW 0.0 0.0 0:00 0 migration/0
3 root RT 0 0 0 0 SW 0.0 0.0 0:00 1 migration/1
4 root RT 0 0 0 0 SW 0.0 0.0 0:00 2 migration/2
5 root RT 0 0 0 0 SW 0.0 0.0 0:00 3 migration/3
6 root 15 0 0 0 0 SW 0.0 0.0 0:03 1 keventd

[root(at)data3 root]# cat < /proc/meminfo
total: used: free: shared: buffers: cached:
Mem: 4223713280 4203782144 19931136 0 37982208 3684573184
Swap: 20859301888 65757184 20793544704
MemTotal: 4124720 kB
MemFree: 19464 kB
MemShared: 0 kB
Buffers: 37092 kB
Cached: 3570800 kB
SwapCached: 27416 kB
Active: 3215984 kB
ActiveAnon: 245576 kB
ActiveCache: 2970408 kB
Inact_dirty: 330796 kB
Inact_laundry: 164256 kB
Inact_clean: 160968 kB
Inact_target: 774400 kB
HighTotal: 3276736 kB
HighFree: 1024 kB
LowTotal: 847984 kB
LowFree: 18440 kB
SwapTotal: 20370412 kB
SwapFree: 20306196 kB

[root(at)data3 root]# cat < /proc/sys/kernel/shmmax
4000000000[root(at)data3 root]# cat < /proc/sys/kernel/shmall
134217728

max_connections = 165
shared_buffers = 32768
sort_mem = 20480
vacuum_mem = 16384
effective_cache_size = 256900

I still in doubt whether this figture is optimized and putting more ram will
help the system throughtput.

Any idea please . My organization is one oof the big hospital in Thailand
Thanks
Amrit
Thailand