[mythtv-users] how to troubleshoot network congestion problems with mythtv

[Tim Coote]

>
>> I seem to have quite a nasty variation in the latency of my home
>> network (ethernet over power, using upa, Comtrend kit). I've got a
>> simple set up with two computers. However, the ping times have a std
>> dev of approx. 2x the mean ping time. With no traffic the ping time
>> is ~4ms, with peaks of 100ms+ every few seconds or so, when
>> streaming live or recorded tv, ping times go up to >20ms (I've not
>> yet measured the variation).
>
> That would not surprise me at all - and my advise would be to invest
> in a real network cable. The cable put in for your power was never
> designed for any form of data transmission and presents a complex and
> changing path - every time a piece of equipment is plugged in,
> unplugged, switched on, or switched off ... the communications path
> alters. That doesn't just mean items on the same power circuit, but
> anything on the same supply phase (if you have more than one) in the
> house.
>
> I strongly suspect these adapters have some significant error
> detection and correction built in - and that could account for the
> occasional high latency. Once you put some traffic on the network
> then I'd expect to see round trip latency go up considerably - you
> have a relatively slow network which is both shared and simplex (ie
> it's like going back to the days before network switches and 100mbps
> or more were common).
>
> Also, some of those powerline network systems are really, really bad
> for the radio environment - and in the UK, the Radio Society of Great
> Britain (RSGB) is having "discussions" with our regulator (OfCom)
> trying to get them to take action against BT who are supplying kit
> that is marked as complying with standards but which wipes out large
> parts of the radio spectrum when used in real applications. Again,
> the power cable isn't designed for data transmission, but it makes a
> fairly effective antenna to leak the signals out.
> Oh look, it was Comtrend kit they complained about :
> http://www.rsgb.org/plt/
> --  
> Simon Hobson
>
Thanks, Simon, but I don't have the option to rewire the house. I've  
read about the ham radio interference, but since the kit is being sold  
by the erstwhile (and possibly current for all I know) policeman of  
spectrum infringement (BT), it seems to me that it's a likely  
technology that mythtv will need to work with.

What I've got to decide is whether to live with no HD, try to fix my  
environment or work to get mythtv to support the variation in latency.  
So I think that I need to know whether my setup's performance is  
usual, and therefore ought really to be handled by the overall mythtv  
architecture, or, unusual, in which case I should change it, or live  
with the consequences.

Looking at the UPA submission to IEEE, streamed TV requires latency of  
<160ms and bandwidth of ~60Mbps. There's no measure of variation of  
either. I have no idea if these are the design assumptions of mythtv.

On your points about variability of performance over domestic wiring,  
I'm surprised that the latency variation is so large (> than the  
average latency to and from google, for instance.) on a moment to  
moment basis (~10-15 seconds gap between the large latency numbers,  
which occur in ones or twos.  I know that it's a simple ethernet  
bridge, but it does seem more stable than, say, wifi.

if no one's got any comparable field measurements, I'm going to have  
to try it at another house to see what happens.

Tim