|NETGEAR ProSafe GS110T Gigabit SmartSwitch|
|Reviews - Featured Reviews: Network|
|Written by Bruce Normann|
|Tuesday, 08 January 2013|
Page 9 of 11
NETGEAR GS110T TestResults
Wired networking equipment is part of a very mature industry. At the very high end of the market, where a router or switch costs $5,000 - $10, 000, there is still a lot of new technology and innovation, however. Recent shifts toward virtualization and cloud computing demand extreme bandwidths and scaling factors. In order to deliver on the many promises the industry has made for these two technologies, something "simple" like a switch or router better not get in the way of overall performance. It's the same way in the home or small business. More and more devices are connecting to smaller, domestic networks, and the backbone has to keep up, or everyone suffers.
One thing about a mature industry is that it's ruled by standards. Given the very nature and mission of networking equipment, it's impossible to imagine that the industry could exist without them. If you want to connect a dozen devices together and have them interface correctly, you need to not just have standards, but adhere to them. So, if a router or switch is rated for 1000BASE-T, it had better perform exactly like the IEEE 802.3-2008 standard says it should. So, let's take a look at several GbE solutions I have in the test lab, and see if there is indeed a standard level of performance from a range of similar products.
In TCP/IP transfers, the NETGEAR GS110T Smart Switch achieved an average speed of 868 Mbps. Compared to the other devices under test, an unmanaged GbE switch and two GbE routers, the GS110T lagged behind in overall speed, by about 5%. The reason is that there were a number of performance dips that occurred consistently enough to harm the average throughput. Here's an example that shows how the transfer rate bounced back and forth between ~915 Mbps and ~830 Mbps. As I said in the testing methodology section, there are myriad reasons why full performance may be hindered or interrupted once you start plugging a number of unique systems together. FWIW, I only saw this kind of behavior on the two network switches I tested; the routers were steady as a rock. Without these dips, just looking at the plateaus, you can see that the potential is there, and it's realized more often than not. Some other process is consuming resources somewhere in the signal chain, or a buffer is getting filled up and then flushed on a semi-regular basis, it's hard to tell which.
The UDP results were much lower on the Smart Switch, reaching only 343 Mbps. The unmanaged switch roughly doubled that performance, and the two GbE routers achieved speeds somewhat close to their TCP/IP performance. On this test, it was the switches that delivered reasonably consistent results and the routers that were all over the chart. The upper traces in the chart below are for the NETGEAR GS110T Smart Switch, and the lower traces are for the Cisco Router. Unlike TCP, the UDP protocol does not divide a message into packets, transmit the packets and then reassemble them at the other end. UDP also doesn't provide sequencing of the data. There's a bit more work for the application programs on each end of the data transfer to do; they have to ensure that the entire message arrived, and that it's in the right order. In this benchmark, the applications were the same for all tests, the PassMark Advanced Networking Test in Client-Server mode. These are all 20 second traces, BTW, and in that time frame each channel of the switch or router is technically capable of shifting approximately 2.5 Gigabytes of data between devices.
The total bandwidth of the GS110T is a full 20 Gbps, the maximum rated capacity of the device, and I wish I had a way of testing that. Ten devices at max speed, in full duplex - I have trouble imagining a scenario that generates that much data, outside of a scientific research lab. Obviously datacom providers and folks like Amazon Web Services, who manage one of the biggest cloud-computing centers in the world, would have no trouble filling this sort of pipeline a thousand times over. For the rest of us, the challenge will almost always rest on the limited capability of a single device on the network, to feed a number of other devices, all at the same time. What happens when everyone decide to back up their data to the NAS at the same time? Four TVs, all streaming HD video from one DVR? Of course, most houses just have one Internet feed, and 3-4 people watching YouTube, Netflix, and Hulu in the evenings.
Now that we've examined the functionality and the performance of the NETGEAR GS110T SmartSwitch, let's look at some of the reasons why you might need one, in our Final Thoughts section.