802.11g read-world throughput?

[Guest]

Archived from groups: alt.internet.wireless (More info?)

Hi,

A quick question: what sort of real-world throughput should I be seeing on
an 802.11g (54Mbps, perfect signal) wireless network? I'm getting less than
10Mbps, which a friend I spoke to said seemed about right, but I was reading
an article today in the computer press where it was stated that they were
getting about double this (20Mbps). Which, if either, is closest to correct?

-dan

 

[fearless]

Archived from groups: alt.internet.wireless (More info?)

"Dan Brill" <news@mangoed.com> wrote in message
news:qELqc.4834862$iA2.564088@news.easynews.com...
> Hi,
>
> A quick question: what sort of real-world throughput should I
be seeing on
> an 802.11g (54Mbps, perfect signal) wireless network? I'm
getting less than
> 10Mbps, which a friend I spoke to said seemed about right, but
I was reading
> an article today in the computer press where it was stated that
they were
> getting about double this (20Mbps). Which, if either, is
closest to correct?
>
> -dan

Is this throughput with or without encryption? AFAIK, turning on
encryption - especially 128 bit WEP - cuts throughput by at least
half, if not more.

More info please.

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

~ A quick question: what sort of real-world throughput should I be seeing on
~ an 802.11g (54Mbps, perfect signal) wireless network? I'm getting less than
~ 10Mbps, which a friend I spoke to said seemed about right, but I was reading
~ an article today in the computer press where it was stated that they were
~ getting about double this (20Mbps). Which, if either, is closest to correct?
~
~ -dan
~

Both are right ... if there are no 11b clients audible from the AP in the
channel, then 20-24Mbps should be possible. If any 11b clients can be heard,
then if the AP is using RTS/CTS protection, then expect 8-9Mbps; if using
CTS-to-self, then expect 12-14Mbps.

Note that these numbers represent maximum aggregate throughput with large
frames. A single stream data transfer between two given computer can of
course yield less due to mistuned TCP stacks, disk I/O bottlenecks or
what have you.

Aaron

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

In article <c8g83t012mi@enews4.newsguy.com>,
Fearless <ima@nottelling.com> wrote:
>"Dan Brill" <news@mangoed.com> wrote in message
>news:qELqc.4834862$iA2.564088@news.easynews.com...
:> A quick question: what sort of real-world throughput should I
:be seeing on
:> an 802.11g (54Mbps, perfect signal) wireless network? I'm
:getting less than
:> 10Mbps, which a friend I spoke to said seemed about right, but
:I was reading
:> an article today in the computer press where it was stated that
:they were
:> getting about double this (20Mbps). Which, if either, is
:closest to correct?

tomsnetworking.com has tests showing a number of devices above 22 Mbps
real throughput. It's a very nice site for hardware reviews -- check
it out, and tomshardware.com as well.


:Is this throughput with or without encryption? AFAIK, turning on
:encryption - especially 128 bit WEP - cuts throughput by at least
:half, if not more.

On -some- devices there are noticable throughput losses when using
WEP or WPA, but not on more modern devices.

--
Usenet is one of those "Good News/Bad News" comedy routines.

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

"Dan Brill" <news@mangoed.com> wrote in message
news:qELqc.4834862$iA2.564088@news.easynews.com...
> Hi,
>
> A quick question: what sort of real-world throughput
should I be seeing on
> an 802.11g (54Mbps, perfect signal) wireless network? I'm
getting less than
> 10Mbps, which a friend I spoke to said seemed about right,
but I was reading
> an article today in the computer press where it was stated
that they were
> getting about double this (20Mbps). Which, if either, is
closest to correct?

on a seemingly "perfect" 108GT connection, I get sustained
rates of around 25-30 megabits/sec.
Wether this is "average" is a moot point ;-)
P.

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

Sorry, it is 64 bit WEP.

I suppose that my next question would be - if this is too low, what could be
going wrong? The Netgear software provides a real-time graph of activity
which suggests that 10Mbps represents about 45% Network Capacity.

-dan

"Fearless" <ima@nottelling.com> wrote in message
news:c8g83t012mi@enews4.newsguy.com...
> "Dan Brill" <news@mangoed.com> wrote in message
> news:qELqc.4834862$iA2.564088@news.easynews.com...
> > Hi,
> >
> > A quick question: what sort of real-world throughput should I
> be seeing on
> > an 802.11g (54Mbps, perfect signal) wireless network? I'm
> getting less than
> > 10Mbps, which a friend I spoke to said seemed about right, but
> I was reading
> > an article today in the computer press where it was stated that
> they were
> > getting about double this (20Mbps). Which, if either, is
> closest to correct?
> >
> > -dan
>
> Is this throughput with or without encryption? AFAIK, turning on
> encryption - especially 128 bit WEP - cuts throughput by at least
> half, if not more.
>
> More info please.
>
>

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

Hi,

Thanks for your reply - I'm not sure what RTS/CTS and CTS-to-self are.
Perhaps you could suggest something which might explain these terms to me?

Certainly, I'm getting between 9 and 10Mbps and there are *no* 11b clients.
I there any easy way to benchmark network performance? I've been using the
free version of Ixia Qcheck to get throughput figures.

To be honest, I've been having so many problems over the last few weeks with
the wireless-networking kit I've purchased (frequent disconnects, crashing
configuration utilities, low speed, hanging data transfers, etc.) that I
wish I'd stuck with my old wired network but it's a bit late now. Still, if
I ever achieve any semblance of real functionality then it'll be worth it!

-dan

"Aaron Leonard" <Aaron@Cisco.COM> wrote in message
news:r0nna05mpmj853998gks1bchci28n643pp@4ax.com...
> ~ A quick question: what sort of real-world throughput should I be seeing
on
> ~ an 802.11g (54Mbps, perfect signal) wireless network? I'm getting less
than
> ~ 10Mbps, which a friend I spoke to said seemed about right, but I was
reading
> ~ an article today in the computer press where it was stated that they
were
> ~ getting about double this (20Mbps). Which, if either, is closest to
correct?
> ~
> ~ -dan
> ~
>
> Both are right ... if there are no 11b clients audible from the AP in the
> channel, then 20-24Mbps should be possible. If any 11b clients can be
heard,
> then if the AP is using RTS/CTS protection, then expect 8-9Mbps; if using
> CTS-to-self, then expect 12-14Mbps.
>
> Note that these numbers represent maximum aggregate throughput with large
> frames. A single stream data transfer between two given computer can of
> course yield less due to mistuned TCP stacks, disk I/O bottlenecks or
> what have you.
>
> Aaron

 

[gary]

Archived from groups: alt.internet.wireless (More info?)

RTS/CTS and CTS-to-self are message sequences used on 802.11g networks when
802.11b clients are present. 802.11b doesn't decode 802.11g frames, so these
short messages (encoded in a way that 802.11b understands) are used to
reserve the network for 802.11g transmissions.

But any 802.11b network overlapping the channel you are using will
significantly degrade you. Check your site survey, be sure you don't see any
APs within 5 channels of you. If you do, change channels.


"Dan Brill" <news@mangoed.com> wrote in message
news:JySqc.4856652$iA2.566229@news.easynews.com...
> Hi,
>
> Thanks for your reply - I'm not sure what RTS/CTS and CTS-to-self are.
> Perhaps you could suggest something which might explain these terms to me?
>
> Certainly, I'm getting between 9 and 10Mbps and there are *no* 11b
clients.
> I there any easy way to benchmark network performance? I've been using the
> free version of Ixia Qcheck to get throughput figures.
>
> To be honest, I've been having so many problems over the last few weeks
with
> the wireless-networking kit I've purchased (frequent disconnects, crashing
> configuration utilities, low speed, hanging data transfers, etc.) that I
> wish I'd stuck with my old wired network but it's a bit late now. Still,
if
> I ever achieve any semblance of real functionality then it'll be worth it!
>
> -dan
>
> "Aaron Leonard" <Aaron@Cisco.COM> wrote in message
> news:r0nna05mpmj853998gks1bchci28n643pp@4ax.com...
> > ~ A quick question: what sort of real-world throughput should I be
seeing
> on
> > ~ an 802.11g (54Mbps, perfect signal) wireless network? I'm getting less
> than
> > ~ 10Mbps, which a friend I spoke to said seemed about right, but I was
> reading
> > ~ an article today in the computer press where it was stated that they
> were
> > ~ getting about double this (20Mbps). Which, if either, is closest to
> correct?
> > ~
> > ~ -dan
> > ~
> >
> > Both are right ... if there are no 11b clients audible from the AP in
the
> > channel, then 20-24Mbps should be possible. If any 11b clients can be
> heard,
> > then if the AP is using RTS/CTS protection, then expect 8-9Mbps; if
using
> > CTS-to-self, then expect 12-14Mbps.
> >
> > Note that these numbers represent maximum aggregate throughput with
large
> > frames. A single stream data transfer between two given computer can of
> > course yield less due to mistuned TCP stacks, disk I/O bottlenecks or
> > what have you.
> >
> > Aaron
>
>
>

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

There are no other wireless networks of any variety (at least that I can
scan for) in the vicinity. Strangely, and this is only today, I'm also
suffering from extremely poor signal quality (first time I've seen this) -
reported as being under 50% by the NIC's configuration utility (though it
still claims that the operating speed is 54Mbps). It seems bizarre since the
AP is less than 5 feet from the NIC. However, this may or may not be related
to any of the problems I've been experiencing since the signal is usually
rated 'very good' (80%+) and the throughput is still extremely low.

I really appreciate your response and efforts to assist. I obviously know
very little about this subject area and Netgear's product support has been
entirely unresponsive (beyond the usual computer-generated email suggesting
a number of possible fixes, which aren't).

-dan

"gary" <pleasenospam@sbcglobal.net> wrote in message
news:EHSqc.541$4l5.118@newssvr22.news.prodigy.com...
> RTS/CTS and CTS-to-self are message sequences used on 802.11g networks
when
> 802.11b clients are present. 802.11b doesn't decode 802.11g frames, so
these
> short messages (encoded in a way that 802.11b understands) are used to
> reserve the network for 802.11g transmissions.
>
> But any 802.11b network overlapping the channel you are using will
> significantly degrade you. Check your site survey, be sure you don't see
any
> APs within 5 channels of you. If you do, change channels.

 

[frank]

Archived from groups: alt.internet.wireless (More info?)

> Strangely, and this is only today, I'm also suffering from
> extremely poor signal quality (first time I've seen this) -
> reported as being under 50% by the NIC's configuration
> utility

> It seems bizarre since the AP is less than 5 feet from the NIC

It IS bizarre! 5 feet? You should be at max signal. Something's rotten in
Denmark.

-Frank

 

[blh]

Archived from groups: alt.internet.wireless (More info?)

"Dan Brill" <news@mangoed.com> wrote in message news:<JySqc.4856652$iA2.566229@news.easynews.com>...
> Hi,
>
> Thanks for your reply - I'm not sure what RTS/CTS and CTS-to-self are.
> Perhaps you could suggest something which might explain these terms to me?
>
> Certainly, I'm getting between 9 and 10Mbps and there are *no* 11b clients.
> I there any easy way to benchmark network performance? I've been using the
> free version of Ixia Qcheck to get throughput figures.
>
> To be honest, I've been having so many problems over the last few weeks with
> the wireless-networking kit I've purchased (frequent disconnects, crashing
> configuration utilities, low speed, hanging data transfers, etc.) that I
> wish I'd stuck with my old wired network but it's a bit late now. Still, if
> I ever achieve any semblance of real functionality then it'll be worth it!
>
> -dan
>
> "Aaron Leonard" <Aaron@Cisco.COM> wrote in message
> news:r0nna05mpmj853998gks1bchci28n643pp@4ax.com...
> > ~ A quick question: what sort of real-world throughput should I be seeing
> on
> > ~ an 802.11g (54Mbps, perfect signal) wireless network? I'm getting less
> than
> > ~ 10Mbps, which a friend I spoke to said seemed about right, but I was
> reading
> > ~ an article today in the computer press where it was stated that they
> were
> > ~ getting about double this (20Mbps). Which, if either, is closest to
> correct?
> > ~
> > ~ -dan
> > ~
> >
> > Both are right ... if there are no 11b clients audible from the AP in the
> > channel, then 20-24Mbps should be possible. If any 11b clients can be
> heard,
> > then if the AP is using RTS/CTS protection, then expect 8-9Mbps; if using
> > CTS-to-self, then expect 12-14Mbps.
> >
> > Note that these numbers represent maximum aggregate throughput with large
> > frames. A single stream data transfer between two given computer can of
> > course yield less due to mistuned TCP stacks, disk I/O bottlenecks or
> > what have you.
> >
> > Aaron

Have a look at this document:

http://www.atheros.com/pt/Methodology_Testing_WLAN_Chariot.pdf

Some interesting facts & figures

BH

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

> It IS bizarre! 5 feet? You should be at max signal. Something's rotten
in
> Denmark.

Well, I may not know a hatchet from a handsaw when it comes to wireless
networking, but I think so too! Anyway, signal strength is back to 72% now.

In addition to benchmarking software for the connection, I'd also like to
see something which counted the number of times the connection died or was
dropped (if only briefly) as this seems to happen quite a lot.
Unfortunately, the Windows Connection Status applet in the notification area
doesn't seem to recognise these brief lapses in connectivity.

-dan

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

> In addition to benchmarking software for the connection, I'd also like to
> see something which counted the number of times the connection died or was
> dropped (if only briefly) as this seems to happen quite a lot.
>

might use boingo to watch for connection drops,,, if has a connection
quality graph window that probably would give you the past half hr or so of
connectivity %'s if thats good enough.

my connection G normal is around 17.5mbit/s with 128wep and about 30' across
the room from the AP/router (di624a)... best i've seen it is around 21. Wep
be it 128 or 64 is only around a 10% drop in through-put. Im on a winME
laptop mostly ... which 17-18 fits right in with most reports... winXP
normal G is 22-25 at best. OppSystem makes a difference for some reason -
maybe someone will enlighten me as why?

Just went from dlink G650 pc card at 72-80% signal to 93-100% with a Compex
G card and blade antenna - its 86-93% w/o antenna attached Big signal &
link quality Kick's!

 

[Guest]

Archived from groups: alt.internet.wireless (More info?)

Taking a moment's reflection, Dan Brill mused:
|
| There are no other wireless networks of any variety (at least that I can
| scan for) in the vicinity. Strangely, and this is only today, I'm also
| suffering from extremely poor signal quality (first time I've seen this) -
| reported as being under 50% by the NIC's configuration utility (though it
| still claims that the operating speed is 54Mbps). It seems bizarre since
| the AP is less than 5 feet from the NIC.

You may be getting some interference from some other source. I would
recommend changing the channel in your router/AP. If you are on the default
of 6, try changing it to 1. If that doesn't help, try channel 11. I was
using channel 1 on my WLAN, and was having connectivity and throughput
issues. I changed to 11, and they went away.