Is 8–9 MB/s on a 5Ghz band (dedicated N) a decent speed?

[theboyk]

I just received a DIR-825 and decided to do some basic speed testing with it. I brought the DIR-825 into a space where I designed/built/manage the network (115+ gigabit drops, all Cat5e [all under max range], cisco switches, cisco router, 3 x Xserves, etc.) — the network peaks at just under 100 MB/s, which is the result of the HDs being used and other overhead — even then, roughly 95 MB/s (about 85 MB/s on average) for a gigabit network isn't too shabby, IMO.

So, I wanted to test the wireless speed of the DIR-825 in this environment. Again, these aren't high-tech tests, just some quick testing. This is what I was using:

Server (hosting the data) = latest Xserve with 1TB RAID1 drive clocking in at 100+ MB/s read/write
Main Switch = stack of Cisco 48 port gigabit managed switches
Computer (receiving the data) = Mac Pro with 1TB RAID1 drive clocking in at 100+ MB/s read/write
File sizes for Testing = 2 – 2.5 GB file sizes

I added the DIR-825 to the network by unplugging my Mac Pro and using that Cat5e cable to hardwire the DIR-825 into the Cisco switch. I then moved the Mac Pro about 15 feet away, in direct line of site with the DIR-825 and enabled wireless N (on the Mac Pro).

I had the DIR-825 set up with the 2.4 Ghz band disabled and the 5 Ghz band enabled. The 5 Ghz band is dedicated to wireless N only, the channel width is 20 Mhz, auto-channels selection. The Mac Pro showed a "Transmission Rate" between 115 – 130 with an RSSI of -18 — pretty strong signal. After running the test 10x I averaged 8.7 MB/s and peaked at 9.3 MB/s. I was about 15 feet away, direct line of site. FYI, at home, running the same sort of test, but going from one room to the room directly 1 floor down (old, hardwood floor), I average between 7–8 MB/s (though, it's peaked at 12.6 MB/s).

So, my question — are these speeds, on average, under these conditions, OK?

Just an aside — I figured I might as well test the wired speed of the DIR-825 so I attached the Mac Pro to the DIR-825 and ran the same sort of tests, but "plugged in" instead of wireless. Again, on this network I get about 85+ MB/s transfer speeds, so I expected the numbers to be drastically lower when I introduced the DIR-825 into the wired environment, but they weren't — I was getting very good speeds, on average 80+ MB/s with the peak hitting 90.6 MB/s. So, IMO, the DIR-825 held its own with respect to wired gigabit speeds.

I'm receiving a Netgear WNDR3700 this afternoon and will hopefully be able to sneak the same tests in.

Thanks,
Kristin.

Note: all of these tests were done in the middle of the day, in an "in-use" production environment as I was looking at the speeds I'd get if I were to actually introduce the device into such an environment.

 

[tipstir]

You know your range and speed will vary. What you get not going to be the same I get? But it gives you idea what type of environment the unit in question can perform and handle. Currently I am testing for EnGenius ESR-9850 Gig 300 Wireless since that is second rated on Tim's SmallerNetBuilder wired/wireless charts I am not going to buy anything like most will do. I ready have a stock pile of routers some are listed below. I would like to get my hands on E3000 but if that's the same as the WRT610N I might pass on it.

By-the-way don't your servers have PCI-X NiCs in them?

 

[theboyk]

Yea, I realize speeds are going to vary depending on the situation, but I'm just wondering if the speeds I'm getting are good or bad for a real-world speed?

And as for the servers/NiC - I'm using the stock gigabit Ethernet ports on my Xserves - why?

Regards,
Kristin.

 

[tipstir]

What are you using to measure the LAN to LAN over Gig. Are those Cisco 3750 48-port.

 

[stevech]

After running the test 10x I averaged 8.7 MB/s and peaked at 9.3 MB/s. I was about 15 feet away, direct line of site. FYI, at home, running the same sort of test, but going from one room to the room directly 1 floor down (old, hardwood floor), I average between 7–8 MB/s (though, it's peaked at 12.6 MB/s).

In 802.11a/b/g/n, the speed is governed by these things, in priority order:

received signal strength, at the client PC *and* at the access point/router. Few routers/APs show you the per-client received signal strength, but it's important, as usually, the client device has lower transmitter power than does the access device.

mode: 11a/b/g/n
In 11n, channel pair bonding a.k.a. 40MHz mode. Is rarely viable.

competition for air time on whatever channel you've chosen, versus other nearby and busy WiFi systems. Emphasis on *busy*.

Other interference in 2.4GHz or 5.8GHz. Certain cordless phones, etc.

Incompatible 802.11 RF modes at the client versus the access device.

BUT also, the test tools you use color things a great deal. QCheck, freeware, is a good quickie to use.
-----------
With an ideal signal strength and no interference or competition for air time...
11g should produce 22Mbps net. This is if the RF mode is 11g at 54Mbps air link speed.

 

[thiggins]

Yes, those speeds are typical of what you'll see from Draft N products at best-case (same room) range and using 20 MHz bandwidth.

You can get an overview of speed test results at all test locations, modes and bands by going to the Wireless Charts, selecting the product(s) of interest and hitting the Performance Table button. The results there are in Mbps.