Right now, my wireless connection is at "Low", 11 mbps.  We are usually at
"Good", around 24 mbps.

Does 11 mbps translate to 1.375 mBps (8 bit byte)?  Does 24 mbps translate
to 3 mBps?

Basically, trying to figure out if we should go to a 1 mB service or 3 mB
service with Charter (currently on an old 384 kB service).

Thanks.

CJ

11th commandment.  Thou shalt not abrev.

Others have answered your question about abreviations.  I'll answer
the question that I think you meant to ask, which is how fast can you
move data at the various speeds.

A 24Mbit/sec OFDM wireless association or connection, will yield about
12Mbits/sec TCP thruput after dealing with all the overhead, sliding
windows, 802.3 enacapsulation, timing delays, and management frames.
An 11Mbit/sec CCK connection will yield about 4-5.5Mbits/sec thruput.

Both wireless speeds are faster than your 1.5 or 3Mbit/sec DSL
service, which also has overhead an inefficiencies.  You'll get about
1.2Mbits/sec thruput with 1500/256Kbits/sec service, and about
2.2Mbits/sec thruput with 3000/384Kbits/sec service.

Personally, I would get the 3Mbit/sec service because it usually
offers higher upload speeds, which is important if you're running
servers, playing VoIP, or doing the VPN thing.

On Sat, 05 Feb 2005 12:44:40 -0800, Jeff Liebermann
<je***@comix.santa-cruz.ca.us> wrote:

I have a wired/wireless (802.11b) setup on Comcast Cable.  The wired
part shows 100.0 Mbps connection on the wired side and when using my
laptop/pda on the wireless side it typically shows 11Mbps.  Confusing
for me is that when using the wireless side with either laptop or pda
it really doesn't appear to be much slower than the desktops on the
wired side.  Maybe I'm just not observing as critically as I should.

Are these figures simply "marketing" numbers?  I assume these numbers
should actually vary with the load on the cable system.

How or where can I actually read "real" xfer values...(Win-XP Home
SP2)?

TIA

Chap <***@*****.***> wrote in news:c2lc01hnl9tigoooj85rt2cata5cg82sjg@
4ax.com:

Your wired connection maybe 100mbps, but the actual bandwidth of the pipe
is 1 - 3Mbit depending on the plan (or if you're lucky like me, 5Mbit/s on
Rogers :)

On Sun, 06 Feb 2005 09:39:49 -0800, Chap <***@*****.***> wrote:

Comcast is usually 3Mbits/sec.  You'll get about 2.5Mbits/sec thruput.

The 100Mbits/sec ethernet speed is the connection speed.  If you've
got two decent computahs, connected on a LAN through an ethernet
switch, you'll get about 85Mbits/sec thruput when copying files
between the two computahs.

Again, that's the connection speed.  You'll get about 4-5.5Mbits/sec
thruput.

Maybe some benchmarking would be appropriate.  Since you have Windoze
XP on the laptop, fire up the "Task Manager" and select the
"Networking" tab.  It should show the network preformance numbers.
http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/taskman_whats_there_w.mspx
There are also a bunch of performance meters avaiable for download.  I
use SNMP to get my numbers, but that's probably overkill for this
test.  Try "IPerf":
  http://dast.nlanr.net/Projects/Iperf/
If all else fails, find a big file and use a stopwatch to time how
long it takes to copy over the network.  Then, do the math.

Nope.  They're for real.  It's kinda like buying a car with a 120mph
speedometer.  It might go 120mph under totally ideal conditions, but
reality has a bad habit of lowering results.  In the wireless case,
the problem is that there is considerable overhead and variables.
Besides congestion and constipation on the cable system, there are
other things that will slow down wireless traffic.  For example
(ignoring slowdowns caused by applications):
  Signal quality (i.e. S/N ratio).
  CCK vs OFDM modulation
  Encryption
  Encryption level
  Router filters
  MAC address filters
  802.3 encapsulation
  RF interference
  RF reflections and multipath
  Co-channel users
  Use of a repeater or WDS repeater.
  Hubs instead of switches
  Slow computah
  Whatever else I forgot.
Every one of these will slow things down from the mythical 11Mbits/sec
to a lower number.  The exact amount of reduction varies with
implimentation, severity (in the case of interference and multipath),
and signal quality.  A few are optional, but most are all part of the
802.11 puzzle.

"Task Manager" -> Networking -> Ethernet

Jeff Liebermann <je***@comix.santa-cruz.ca.us> wrote:
Iperf requires two ends.  If you have two computers on your network, you
can check that performance.  If you have a Unix account at an ISP, you can
load iperf there.  Default NAT router and firewalls will let you test
upload.  You'd have to map a port through the firewall to test download.

http://www.dslreports.com/stest will test up and down to a couple of sites
on the internet.  The speeds seem pretty stable and respectable.

There ya go.  Copying between two wireless computers will also show off
another attribute of Wireless that is important.

That's not very exciting.  I don't think I believe it.
start-run-perfmon.msc
    + Performance Object = Network
Numbers agree with dslreports.
    + Performance Object = TCP
shows some counters that might be errors.

On Mon, 7 Feb 2005 03:52:33 +0000 (UTC), d***@XReXXBasic.usenet.us.com
wrote:

That will work for his internet download and upload speed.  However,
he wanted the speed of his 802.11b wireless connection, not the
internet connection.  The internet connection will be limited to about
3Mbits/sec, while the 11Mbit/sec wireless will be somewhat faster.
The only way to get a real number for 802.11b is to do it locally and
not through the internet.  However, I forgot that he only has one
computah, which will make using Iperf difficult.

Agreed.  It's the best.  However, every once in a while, I find that
it's either overloaded, or yielding bizarre results.  So, I go hunting
for an internet speed test at:
  http://www.testmyspeed.com
which offers about 30 test sites for California.  For SBC DSL tests, I
use:
  http://support.sbcglobal.net/dsl/speedtest/
I'm not sure if Comcast has a local speed test tool.  Methinks not,
but I'm not sure.

Instead of Iperf, methinks maybe something more "windowish" might be
appropriate like "Netstat Live":
  http://www.analogx.com/contents/download/network/nsl.htm
I played with it a few times, got some funny numbers on gigabit LAN
connections, and went back to SNMP.  It will probably work well enough
for this test.  I just installed it on my W2K system.  No smoke or
fire, yet.

OK.  Better than the Task Manager thing but make ugly graphs.  (Can
you tell that I don't use these much?)  I prefer SNMP for everything
using Net-SNMP for grabbing numbers or MRTG for making pretty graphs.
http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/sag_snmpchecklist.mspx
However, methinks this might be a major challenge for a beginner, so I
won't suggest using SNMP.

Jeff Liebermann <je***@comix.santa-cruz.ca.us> wrote:
Or any other test of wireless speed.  Gotta have two ends, regardless of
the test.  What I was pointing out is that it is unlikely that most folks
would have access to a remote iperf client/server.  If you have a two
systems on your network, preferably with one wired and one wireless, you
can see the speed.

You mentioned the stopwatch and big file trick.
Most people have access to a "personal web site" on mchsi.
That allows ftp upload and download for clocking the internet speed, and
might be local to your ISP connection point, maybe not.

Closer is nicer.  People should use their own ISP's speed test, if there is
one.  Some of the ISP-internal tests seem strangely limited.  I wonder if
they work better for people on that network, and are deliberately hampered
going to the internet. 

The "current bandwidth" in perfmon bounces around on my 802.11g connection,
almost a signal strength indicator ;-)
Maximum 54000000, Minimum 24000000.

One odd thing that I noticed.  If I run through a VPN, my measured upload
speeds are higher than without VPN.  I thought my cable was capped to 256Kb
upload, and that has seemed correct over time, with my upload being
measured around 234Kb.  With VPN, I see 700Kb to the test sites, and a
1953Kb iperf to a server that is colo with the VPN endpoint.  I don't know
at all what that means.  I haven't paid much attention to ftp uploads
through VPN.  I usually do those "inside".

http://192.168.100.1/signal.html shows my cable modem has
Downstream    Frequency 561000000 Hz Locked
Upstream    Frequency 23984000 Hz Ranged
Does that mean anything for speed?

On Mon, 7 Feb 2005 19:31:49 +0000 (UTC), d***@XReXXBasic.usenet.us.com
wrote:

Weird but really interesting.  My guess(tm) is that your cable ISP may
have had some "issues" with commerical customers using VPN's and have
implimented some form of QoS or bandwidth management by port number to
make it work better.  I do the same with VoIP ports.  If that were
true, they would have a rather substantial performance edge selling to
corporate clients. 

Nope.  Those are the cable RF frequencies used by the modem.  561MHz
is roughly cable channel 81.  23.984MHz is the return channel, which
is usually between 5-42MHz.
  http://www.jneuhaus.com/fccindex/cablech.html

Jeff Liebermann wrote:
d***@XReXXBasic.usenet.us.com
don't know
Its more likely that your cable company has a "cheap" bandwidth
management
device that works on straight IP but not with vpn encapsulations, so
you're actually working around it.

Dennis
Emerging Technologies Bandwidth Management and traffic shapers

Linksys or Netgear?
D-Link Blues: USB adapter connects to router, no internet access
Wfi over water
Hi.. setting up wireless ISP
Cisco 1231G WPA2 (AES)
Boost WiFi Signal?
Linksys Firmware
yet another netgear problem
Linksys Wireless Network Monitor
Netgear WG511v2 can't connect