Router Performance

I don't have multiple broadband connections (hell, I don't even have one!), so, as usual, I ran all tests on my test LAN using Qcheck in various configurations. (Details of how I use Qcheck for router testing can be found here.)  But since the PRO800T has dual WAN capability, I had to create some additional tests which are described below.

Test Setup 1: WAN Port 1: "Normal". Wan Port 2: "Off". One set of Qcheck test partners

Comments: Transfer rate is average for present generation routers, and plenty fast for most broadband connections.  UDP streaming performance was solid, with no lockups.  Also saw no speed difference with or without DMZ enabled.

Test Setup 2: Both WAN ports: "Normal". Load Balance:50% (default). One set of Qcheck test partners

I switched Qcheck between the WAN ports and found equal performance on both ports with numbers pretty much the same as Test Setup 1. No hiccups when I switched Qcheck from WAN port to WAN port.

Test Setup 3: Both WAN ports: "Normal". Load Balance:50% (default). Two sets of Qcheck test partners

Both WAN ports were plugged into my SMC7008BR Barricade's LAN ports which served them DHCP addresses of 192.168.3.115 and 192.168.3.116.

One Qcheck console was run on a PRO800T client with IP 192.168.0.3, aimed at a test partner of IP 192.168.3.181. Both test partners had 10Mbps/Half duplex connections.

The second Qcheck console was run on a PRO800T client with IP 192.168.0.2, aimed at a test partner of IP 192.168.3.108. Both test partners had 100Mbps/Full duplex connections.

I ran Qcheck Transfer Rate tests with a 1MByte file size from LAN to WAN. The LAN to WAN tests were used so that I didn't need to worry about Port Forwarding or DMZ in order for Qcheck to work.

The test results and the LEDs on the PRO800Ts front panel indicated that the transfers between the two test partner pairs weren't always being sent out on separate WAN ports. Even when I started one test about a second after the other, sometimes the PRO800T would use the same port, and sometimes it would use two ports.

QCheck Tests
Test Description	10Mpbs / Half partners Qcheck Transfer Rate (Mbps)	100Mpbs / Full partners Qcheck Transfer Rate (Mbps)
Solo	3.5	3.6
Simul - 1 WAN	1.8	1.7
Simul - 2 WAN	2.6	2.6

Comments: These results show behavior that pretty much matches Nexland's description.  The Solo case runs one test partner pair at a time.  The Simul - 1 WAN case runs both test partner pairs through one WAN port at the same time, with the result of one WAN port's worth of bandwidth being shared between the test pairs - a 50% reduction in bandwidth for each test pair.  The Simul - 2 WAN case, which runs both test partner pairs through the 2 WAN ports at the same time, doesn't double bandwidth to the 7Mbps you might expect, but still yields a 50% improvement in total bandwidth vs. the Solo case.

Test Setup 4: Both WAN ports: "Normal". Load Balance:50% (default). Two sets of test partners doing HTTP download via web browser.

As an alternative to Qcheck, I also ran the HTTP download test (described here) with the two previously described test partner pairs instead of using Qcheck.

While running this test I found through experimentation that I had to wait about 2 seconds after starting the first download to start the second download or the PRO800T would use the same WAN port for both transfers.  Here are the results:

HTTP Download Tests
Test Description	10Mpbs / Half partners Qcheck Transfer Rate (Mbps)	100Mpbs / Full partners Qcheck Transfer Rate (Mbps)
Solo	4.4	6.1
Simul - 1 WAN	3.1	3.7
Simul - 2 WAN	3.6	4.7

Comments: These results are harder to interpret than the Qcheck results, but you can see the same general trend between the three test types.  The percentages don't come out as cleanly however and the total bandwidth using both WAN ports seems to come out to more than 8 Mbps vs. the 7Mbps indicated by the Qcheck tests.  You also still see some overhead in the Simul - 2 WAN tests that eats up some bandwidth.

Summary

The PRO800Turbo looks like it performs pretty much as advertised and that's a pretty impressive accomplishment for a product at a price point of about $400!

But who has two Broadband connections anyway?!  From the mail that I get, there are some individuals who have them and who have been waiting for a product like the PRO800Turbo.  However, I'd expect that its broader use will be by small to medium sized businesses with medium sized LANs.  These folks usually have enough users to outgrow a single cable or DSL connection and the PRO800Turbo will give them an alternative to more expensive WAN connection options, as well as give them a more reliable connection via its auto fail over capability.  

So if you've got a couple of Broadband connections lying around and want to really get your money's worth out of them, you should take advantage of Nexland's 30 day money-back guarantee and give the PRO800Turbo your own torture test!

(By the way,  if you don't need the second WAN port, but do need the extra LAN ports, the non turbo PRO800 is also available for $100 less.)