Re: Wireless considerably faster one way than the other
- From: Jeff Liebermann <jeffl@xxxxxxxxxxxxxxxxxxxxxx>
- Date: Fri, 16 Jun 2006 09:51:37 -0700
"Dave Brown" <softwaredave@xxxxxxxxx> hath wroth:
Acer Travelmate laptop with built-in Intel 2200GB 802.11b/g wireless
network adaptor, XP Pro.
Netgear ME102 802.11b Access Point, plugged into network hub.
Is the ME102 running the Netgear firmware, or has someone installed
firmware from a DWL-900AP+ or WAP11?
2 more desktop machines plugged into the hub, both Win 2k, it's a 100Mb
network.
The Access Point is approximately 3 inches away from the laptop.
Too close. The receivers will overload at that distance. There are
also some timing problems that appear at such a short distance. I'm
not sure if there's a real minimum distance, but methinks about 12" or
more is safe.
I'm experiencing really bad network throughput at the laptop end, but
have discovered that copying a large file from the laptop to one of the
desktop machines is fast - the problem seems to be only when
transmitting data TO the laptop. I get all sorts of network errors,
including disconnections. Signal strength on the laptop is a full set
of green bars when I look at the connection status, "excellent" when I
hover over the tray icon.
Yep, been there. I've seen this when the 10baseT-HDX (half duplex)
interface on the access point ends up with a different protocol than
your network hub or switch. In particular, with a 10/100 hub or one
of the Netgear FS-xxx series 10/100baseT switches. This does not
cause disconnects, so that's probably not the problem.
If you're using a 10/100 hub (not switch), I would certainly want to
try using a 10baseT only hub, or a 10/100baseT switch, as the buffers
in the dual speed hubs are always a problem. Going from 10 -> 100 is
not a problem because the buffer can be emptied faster than the 10
port can fill it. However, going from 100 to 10 requires that flow
control and buffering work overtime. That's the way you get
assymetrical performance.
I downloaded QCheck, installed it on a desktop machine and the laptop.
When I run a throughput test at the laptop, sending from the laptop to
a desktop machine with 100kb packet size, I get roughly 5 to 6 Mbps,
which as I understand is as good as you can expect from 802.11b.
Maximum theoretical speed at 10Mbits/sec is 7.1Mbits/sec for UDP and
5.9Mbits/sec TCP. Most benchmark tests use file transfers which are
TCP. You're going as fast as 802.11b will go.
However, when I run the test from a desktop machine to the laptop, I
get around 30 kbps. Yep, kilobytes per second. And that's when it
works, most of the time the test fails - it only works reliably when I
reduce the packet size down to, say 5kb, then it shows a throughput of
about 1 Mbps.
Fine. Take the other potential sources of error out of the picture.
Plug the desktop machine directly into the ME102 access point and try
again. You might need a crossover ethernet cable.
Anyone have any idea what could be causing this? I've tried all
possible channel settings, and combinations of all other settings I can
see.
--
Jeff Liebermann jeffl@xxxxxxxxxxxxxxxxxxxxxx
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
.
- References:
- Wireless considerably faster one way than the other
- From: Dave Brown
- Wireless considerably faster one way than the other
- Prev by Date: Re: WiFi Range Extension
- Next by Date: Re: Linksys WAP54G bridge mode and WPA
- Previous by thread: Re: Wireless considerably faster one way than the other
- Next by thread: Re: Wireless considerably faster one way than the other
- Index(es):
Relevant Pages
|
