My first reaction is, if the applications you are running are that network intensive, then the quality of the switch should have stats available to tell you....
Afterall, ,just because it is a switch with 100Mb/s interfaces does not mean that it can actually process packets
- at a rate of 100Mb/s devided by the MTU of the network (8 x 1500 bytes)
- on all interfaces at the same time at that rate (8 port, therefore 4 in/4 out at the above rate).
so just moving to a gigabit interfaced switch may not be the 'simple' answer anyway.
I suspect you need to see if the peaks of demand in the network are causing your significant bottlenecks... would being 10x faster allow your equipment to use that extra speed... i.e just because I am on an expressway, does not mean my car can go that fast... sorry, I kinda said that already
- does it matter if you now take 0.1 sec instead of 1 second to access a file ? are other network waits not more dominant ?
- are you seeing significant/continuous collissions (a sign of peak overload, note that there will always be some by design)
being more practical...
What type of switch is it ?
does it have snmp support, that would allow you to connect stats S/w that would provide you with the picture you really need.. second by second, min by min throughput and by interface (if poss). so you can see the perfornce over the hour/day as a graph.
I suggest a guick google for 'network monitoring software' and 'traffic monitoring software'
prtg is one of the frequent recomended tools, but there are others...