Page 2:What Makes A Gigabit Network? Cards, Cables, And Hubs
Page 3:First Test: How Fast Is Gigabit Supposed To Be, Anyway?
Page 4:Network Speed Limiting Factors
Page 5:Test Systems
Page 6:Network Tests: Setting Our Expectations
Page 7:Network Tests: Are We Getting Gigabit Performance?
Page 8:Testing-Cabling Factors
With our tests delivering performance so close to ideal, we’re probably not going to see any appreciable differences when we change a few cable variables. But we wanted to try it anyway just for curiosity’s sake, mostly to see if it would bring our results any closer to the theoretical performance speed limit.
We did four tests:
Test 1: Default
In this test, we used two 25 ft. cables, each running from one computer to a gigabit switch. We left the cables as they sat, running right beside power cables and power bars.
Test 2: Remove Power Cable Interference
This time, we used the same 25 ft. Cat 5e cables as in the first test, but we moved the network cable as far as possible from power cables and power bars.
Test 3: Shorten Cable length from 50 ft. to 28 ft.
In this test, we removed one of the 25 ft. cables and replaced it with a 3 ft. Cat 5e cable.
Test 4: Replace Cat 5e cables with Cat 6 cables
In this final test, we replaced the 28 ft. Cat 5e cables with 28 ft. Cat 6 cables.
In short, our cable testing didn't show us a lot of difference anywhere, but we did learn a few things:
Test 2: Running Cables Close to Power Cables:
In a small LAN setup like ours, our tests showed you don't have too much to worry about if you have to run your network cable close to a power bar or two. While it's not ideal, it's probably not going to have an appreciable effect on network speeds. Having said that, it should be avoided if possible, and you should keep in mind that these results were taken from a small home network.
Test 3: Shorter Cable Length
This wasn't much of a test, but was just something we tried to see if there was an appreciable difference. We have to remember that when replacing the 25 ft. cable with a three ft. cable that any difference we might see is not necessarily a result of the shorter distance, but may indicate a cable problem. In any case, we didn't really see any differences of note in most tests, although we did see an abnormal jump in performance between the client-to-server C: drive file copy.
Test 4: Cat 6 Cables vs. Cat 5e Cables
Once again, we didn't see much difference here at all. Since we're dealing with 28 ft. of length, we might see a difference over longer cabling. But as long as the cables are in good shape, it looks like Cat 5e cables will work just as well in a home gigabit network with 50 ft. of cable between the peers.
It's also interesting to note that these cabling variables had absolutely no effect on transferring data from one RAM drive to another. Clearly, something else in the system is bottlenecking network performance at the magic 111 MB/s number; as we mentioned, there is some network overhead to be expected so this result is perfectly acceptable.