Besides VLANs, the other thing that pushes folks to smart switches is the need for QoS. I think the GS108T-200 and GS110TP take a bit of a step back here. Configuring queues and priorities is probably fun for real switch geeks. But most of us just want a quick-and-dirty way to choke down the pipe for our favorite bandwidth hogs. Figure 7 shows the simple per-port Ingress / egress rates that can be set on the original GS108T that allow doing just that.
Figure 7: GS108T (original) QoS Rate Control
All you need to do is decide whether you want to control the rate of traffic coming into (Ingress Rate) or out of (Egress Rate) a port, select the desired rate from the dropdown and save the settings. The main limitation of the GS108T's rate control is that it may not be much help for limiting Internet traffic because there aren't a lot of "clicks" below 10 Mbps and only one below 1 Mbps. But if you need to limit traffic on your LAN, the settings are probably fine.
In contrast, Figure 8 shows the GS108T-200 and GS110TP's equivalent function, the Interface Shaping Rate.
Figure 8: GS110TP CoS configuration
Instead of a simple bandwidth pick list, the Interface Shaping Rate field is used to set the maximum port or LAG egress bandwidth, in increments of 64 kbps. A value of 0 means no bandwidth limit is applied. I don't know why the allowable range is 16 to 16384. A pick list like the old GS108T would have been much simpler and clearer.
Another gotcha is that there is no ingress limiting. When I raised this issue with NETGEAR, they said the ingress rate limit is not as useful because it isn't as accurate as egress rate, especially with TCP/IP traffic. According to NETGEAR, the bursty nature of Ethernet traffic can cause packets to be dropped (over a long time period) even if the overall traffic rate is under the rate limit. Egress rate, on the other hand, can be controlled by buffering packets until the set buffer size is met. And if you really want ingress limiting, you can do it with Diffserv policies.
I don't know about you, but even looking at the DiffServ screens is daunting. So until NETGEAR provides some sort of DiffServ wizard, I'll have to forego simple ingress bandwidth limiting on the GS108T-200 and GS110TP.
I don't test switches for performance because any test I can do with two computers is going to test the computers' throughput more than the switch under test. But for those of you (and you know who you are) who are convinced that you can detect a performance difference among low port count Gigabit switches based on their specs, well, here you go:
|Switching Capacity||Non-Blocking Full WireSpeed on all packet sizes|
|Forwarding method||Store and forward|
|Bandwidth||20 Gbps full duplex|
|Buffer memory||512 KB|
|Address database size: (MAC) addresses per system||4K|
|Latency||Less than 4 us for 64-byte frames|
Table 1: GS110TP Performance Spec Summary
It's too bad that the GS110TP's $260 price is such a jump up from the $100 or so that the GS108T-200 will set you back. But if you're looking for a cheaper PoE alternative, there's always the unmanaged GS108P. For around $150, you get eight unmanaged Gigabit ports and 50W of 802.3af PoE on four of them.
But if your wallet is fatter and you want all the hours of fun and enjoyment that a "smart" switch can bring, plus the convenience of PoE, the GS110TP will provide it all in abundance. And the lifetime warranty is just icing on the cake.