From: Andy Hewitt on
David Empson <dempson(a)actrix.gen.nz> wrote:

> Andy Hewitt <thewildrover(a)me.com> wrote:
>
> > With all my testing recently, I'm very puzzled by discrepancies in
> > speeds between using wired and wireless networking.
> >
> > I realise that wireless issn't as good as wired in the main, however,
> > the disrepancy just can't be explained by that, can it?
> >
> > I have 10/100 Ethernet or a selection of wireless protocols to use.
> > Currently I have my Airport Extreme (10/100 ports only) box setup to use
> > 5Ghz wide (dual channel), which *should* give me 2x 270Mbps, yes?
>
> No, it will give you about 300 Mbps at best. If it wasn't wide band it
> would be limited to more like 130 Mbps.

I take it that's each way per channel?

> You can find out the nominal speed of your client's wirless connection
> by using Airport Utility to examine the list of clients connected to the
> base station. To get this, go into Manual Setup for your base station,
> then choose "Logs and Statistics" from the Base Station menu, and click
> the Wireless Clients tab. Each connected client is identified by MAC
> address (Airport ID).

Yeah, I knew about that one. I have also now found the one suggested by
Jaimie, which is a bit easier to get to :-)

> > I did a quick test with a 200MB file copy to the drive attached to the
> > Extreme box, which is USB2 - 480Mbps. And yes, I know about the
> > limitations of that, and it's only theoretical maximums.
>
> Divide that by 2 to get approximate real world maximum throughput for a
> hard drive connected to a USB port on a computer. Possibly somewhat
> slower for an Airport Extreme.

Righto.

> > The actual speeds are *well* below the capability of the network though.
> > The 200MB file took 4mins 5secs on Airport, and 3mins 8secs on Ethernet.
> > iStat menu only shows a peak activity of about 18Mbps on Airport and
> > 32Mbps on Ethernet.
>
> USB is the bottleneck in the case of Ethernet (32 Mbps is actually
> pretty good for a USB connected hard drive).
>
> Airport has further performance overheads.
>
> Ethernet is almost always full duplex these days: it can simultaneously
> transfer 100 Mbps both ways.
>
> Any form of wireless networking is half duplex - each end must take
> turns transmitting. This limits the maximum throughput due to having to
> stop and wait for the other device to respond. A file transfer involves
> a constant stream of acknowledgement going the other way, which slows
> down the transfer.
>
> That combined with the actual bit rate for your client may explain the
> apparent lack of speed for Airport.
>
> If there are multiple base stations or multiple computers connected to
> the wireless network, the peak performance will be worse.

Yeah, I have three in fact. The Netgear broadband router, which is there
because it's the best one for Emily's iPod to connect to at 'g' speeds.
The Express box, because I use Airtunes. The Extreme box, because it
separates my 'n' networking from Emily's 'g', and it's handy for the
shared printer, and networked drives.

Looking closely at how the files transfer, it looks like there's a lot
of buffering going on, the file transfers for a bit, then pauses, then
transfers a bit more, and so on. It looks like it could complete the
file transfer almost twice as fast if it didn't keep pausing.

--
Andy Hewitt
<http://web.me.com/andrewhewitt1/>
From: Fred McKenzie on
In article <1jk1mwd.9ordk41gg4898N%dempson(a)actrix.gen.nz>,
dempson(a)actrix.gen.nz (David Empson) wrote:

> Any form of wireless networking is half duplex - each end must take
> turns transmitting. This limits the maximum throughput due to having to
> stop and wait for the other device to respond. A file transfer involves
> a constant stream of acknowledgement going the other way, which slows
> down the transfer.
>
> That combined with the actual bit rate for your client may explain the
> apparent lack of speed for Airport.

David-

AirPort is a two step system. Each data packet is sent to the AirPort,
and then the AirPort sends it to the destination. The acknowledgement
returns the same way. Doesn't this divide the theoretical data rate by
2? Or is that factored into the calculation?

Fred
From: Andy Hewitt on
Andy Hewitt <thewildrover(a)me.com> wrote:

> Jaimie Vandenbergh <jaimie(a)sometimes.sessile.org> wrote:
[..]
> > The "line speed" that's in play can be seen on the Mac by alt-clicking
> > the wifi fan in the menubar, "transmit rate". (starts up wifi...)
> > Mine's 218Mbit/sec, between a 2009 iMac and a dual-channel AE 15 feet
> > away through one set of floorboards.
> >
> > It's 162 now, after a renegotiation to get a better signal lock.
> > Reliability wins over speed, it's the Apple way.
>
> Wow, I didn't know about that one. Nice, thanks. That answers an awful
> lot then, mine's only showing 54Mbps. I have a very early model of the
> extreme boxes, is it possible it can't properly manage the 270Mbps
> protocol?
>
> Hmmm.....
>
> Even more 'wow', I switched it back to 2.4GHz, and now get 78Mbps. I
> then tried 'Interference Robustness', and it went down to 7Mbps.

Right. Couldn't let that go, so I've been messing about now. I've
relocated the Airport box from the cupboard it was in, and now have a
Transmit Rate of 243 :-). Much betterer! It pushed the peak data
transfer rate to 44.5Mbps.

--
Andy Hewitt
<http://web.me.com/andrewhewitt1/>
From: Jaimie Vandenbergh on
On Sun, 13 Jun 2010 20:45:00 +0100, thewildrover(a)me.com (Andy Hewitt)
wrote:

>Andy Hewitt <thewildrover(a)me.com> wrote:
>
>> Jaimie Vandenbergh <jaimie(a)sometimes.sessile.org> wrote:
>[..]
>> > The "line speed" that's in play can be seen on the Mac by alt-clicking
>> > the wifi fan in the menubar, "transmit rate". (starts up wifi...)
>> > Mine's 218Mbit/sec, between a 2009 iMac and a dual-channel AE 15 feet
>> > away through one set of floorboards.
>> >
>> > It's 162 now, after a renegotiation to get a better signal lock.
>> > Reliability wins over speed, it's the Apple way.
>>
>> Wow, I didn't know about that one. Nice, thanks. That answers an awful
>> lot then, mine's only showing 54Mbps. I have a very early model of the
>> extreme boxes, is it possible it can't properly manage the 270Mbps
>> protocol?
>>
>> Hmmm.....
>>
>> Even more 'wow', I switched it back to 2.4GHz, and now get 78Mbps. I
>> then tried 'Interference Robustness', and it went down to 7Mbps.
>
>Right. Couldn't let that go, so I've been messing about now. I've
>relocated the Airport box from the cupboard it was in, and now have a
>Transmit Rate of 243 :-). Much betterer! It pushed the peak data
>transfer rate to 44.5Mbps.

I've been glancing at mine through the day, and it's varied from the
numbers up top down as far as 71, now at around 140Mbps. Not suitable
for guaranteed bandwidth!

Cheers - Jaimie
--
To every complex problem there is a solution which is simple, neat and wrong.
-- HL Mencken
From: David Empson on
Andy Hewitt <thewildrover(a)me.com> wrote:

> Jaimie Vandenbergh <jaimie(a)sometimes.sessile.org> wrote:
>
> > On Sun, 13 Jun 2010 10:01:25 +0100, thewildrover(a)me.com (Andy Hewitt)
> > wrote:
> [..]
> > The "line speed" that's in play can be seen on the Mac by alt-clicking
> > the wifi fan in the menubar, "transmit rate". (starts up wifi...)
> > Mine's 218Mbit/sec, between a 2009 iMac and a dual-channel AE 15 feet
> > away through one set of floorboards.
> >
> > It's 162 now, after a renegotiation to get a better signal lock.
> > Reliability wins over speed, it's the Apple way.
>
> Wow, I didn't know about that one. Nice, thanks. That answers an awful
> lot then, mine's only showing 54Mbps. I have a very early model of the
> extreme boxes, is it possible it can't properly manage the 270Mbps
> protocol?
>
> Hmmm.....
>
> Even more 'wow', I switched it back to 2.4GHz, and now get 78Mbps. I
> then tried 'Interference Robustness', and it went down to 7Mbps.

Any walls in the way? Lower frequencies (2.4 GHz) travel better through
solid objects than higher frequencies (5 GHz).

--
David Empson
dempson(a)actrix.gen.nz
First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: DNLA Software
Next: Search & Replace font in Pages 09