[LINK] Advertised NBN plan speeds impossible to achieve: ACCC

[Paul Brooks]

On 12/11/2019 10:32 am, Glen Turner wrote:
> Hi Tom
>
> It *is* widely understood by the general public. That's why the
> definition has retained its popularity.
>
> At the moment if you buy a 100Mbps service and you want to distribute
> that around your house you go to OfficeWorks and buy a 100Mbps "Fast
> Ethernet" switch (or access point, but let's use switch for this
> exposition).

Glen - NBN could always deliver 105 Mbps for the  '100 Mbps' label which would
completely avoid the confusion you refer to.

Many other goods are sold with a little more than the label indicates, often to avoid
the sort of 'I was short-changed' complaints to the ACCC or Fair Trading.

400g cans of beans have more than 400g of beans inside them. 1 litre bottles of
softdrink have slightly more than 1 litre of fluid in them.

A 250g packet of eggnoodles actually weighs 280g excl the packet (I just checked).

In the other direction, every car speedometer I'm aware of indicates around 5% too
high, so that when the needle reads 60 km/h its actually doing 57 km/h or thereabouts
to avoid people being booked for exceeding the limit when they were sticking to the
correct speed as indicated by the car dashboard.

Theres nothing inherent in the network which requires a 100.0 Mbps  layer-0 bitstream
- all NBN internal network is gigabit/10/100Gb or higher, the 25.0/50.0/100.0 Mbps
link speed is purely an artifact of a traffic-shaper algorithm that can be changed in
an instant.  Even the physical Ethernet socket on the NTD is a gigabit interface on
all the NTDs I'm aware of, so providing 105-ish Mbps of layer-2 throughout isn't an
issue there either.

NBNCo could increase their layer-2 throughput for all products by 3 - 7% without
changing the labels, or the understanding at all - and if people are then measuring
51.5 Mbps or 103 Mbps on their Internet speed-tests, they have a lot less to complain
about than if they measure 97, and are told by engineers 'for the 100 Mbps service you
bought, 97 is the maximum you'll ever measure' and try to understand why - or worse,
complain to their RSP they aren't getting what they are paying for, and pushing all
the customer service costs onto the RSPs to explain 'but but but thats all the NBN
delivers to us' and set up another example of finger-pointing.

Paul.


>
> It's readily understandable that if you want to buy a 300Mbps service
> and want to distribute that around your office you go to OfficeWorks
> and buy a 1000Mbps "Gigabit Ethernet" switch.  If people have issues
> with SI units then that's well within the capabilities of OfficeWork's
> staff.
>
> Now let's say the ACCC alter the definition of the Internet service
> speeds, and the client buys a 1.1Mbps service (the description of a
> 100Mbps service which meets the definition of "worst case transport
> layer throughput across the link" which ACCC is proposing, although
> they don't realise it). That OfficeWorks worker would be wholly excused
> when they sold that customer a 10Mbps "Ethernet" switch.  A switch
> which is ten times too slow.
>
> In short, the ACCC is acting suboptimally. Viewing only a small part of
> the consumer experience with networking.
>
> I view the current definition of bandwidth as the "single useful
> rating" which you seek.  It's independent of traffic mix. It informs
> the customer of which matching products they should purchase.  That it
> misstates achievable throughput by a small amount is readily
> understood. A site's technical staff can measure their site's packet
> size distribution and easily calculate the transport-layer throughput
> across the link, should those few percent matter.
>
> -glen
>
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link