General : STM always enforced?

I will see. I bet it will b more tho! Been getting 4 hours lately when doing more than say 6.5GB I did 20 so eek!.

---------- Post added at 00:46 ---------- Previous post was at 00:04 ----------

Yes it's come off.

[Kushan]

we really need a better way of monitoring our data usage. Sadly, I get the feeling Virgin is putting this off as they're not monitoring it correctly themselves!

[Sephiroth]

Someone might want to consider my mathematical hypothesis but lets say that in my segment/node, all I can get at peak time is 20 meg (which happened yesterday) and I'm defo not STM'd.

Doesn't that mean that everyone else on 120 meg can't get more than 20 meg at the same time whether or not STM's?

In that case, if STM was lifted, they'd still be limited to 20 meg due to the saturation?

Anything wrong with that?

[Kushan]

Originally Posted by Sephiroth

Someone might want to consider my mathematical hypothesis but lets say that in my segment/node, all I can get at peak time is 20 meg (which happened yesterday) and I'm defo not STM'd.

Doesn't that mean that everyone else on 120 meg can't get more than 20 meg at the same time whether or not STM's?

In that case, if STM was lifted, they'd still be limited to 20 meg due to the saturation?

Anything wrong with that?

I guess the only thing wrong with that is assuming there's an even distribution of bandwidth and that everyone can only get 20meg. I'm not really sure how the load balancing works, but if you're only getting 20meg, does that definitely mean everyone else is getting only 20meg, or are some getting more than that?
I really do wish we had more data on how effective STM is under heavy congestion.

[Sephiroth]

My point, Kush, is that STM ion a saturated environment makes no speed difference to what the user would otherwise receive. The CMTS just punts stuff out; it can theoretically dynamically change one or more downstream channels according to configuration settings but I've never seen that in normal circumstances in any of the event logs published on the forums. Indeed it would be pointless because peak time on 8 channels would be peak time on the other 8 channels in a service group.

So is my hypothesis wrong?

[qasdfdsaq]

Originally Posted by Kushan

we really need a better way of monitoring our data usage. Sadly, I get the feeling Virgin is putting this off as they're not monitoring it correctly themselves!

Indeed, VM have always been useless at this, with BT being better but not by a whole lot. The mobile networks are really the ones to beat, as they all provide near-realtime data usage counters (some more accurate than others )

Doing it on the user's router would be the most accurate mechanism but can be very resource intensive depending on the level of detail you require, and doing it on the ISP's edge router would require considerably more network integration than they seem bothered to achieve Then again STM by definition requires per subscriber data counters so if they already have that then it's just pulling the data off in the right way and sending it on to appropriate display systems that's needed. Then then again, they did get lazy with the STM turning it down to now polling only once every 15 minutes instead of realtime; not sure what their reasons were but I have a few suspicions...

---------- Post added at 14:50 ---------- Previous post was at 14:47 ----------

Originally Posted by Sephiroth

Someone might want to consider my mathematical hypothesis but lets say that in my segment/node, all I can get at peak time is 20 meg (which happened yesterday) and I'm defo not STM'd.

Doesn't that mean that everyone else on 120 meg can't get more than 20 meg at the same time whether or not STM's?

Not necessarily. Again it'd depend on how well their load balancing is set up and the specific metrics used. STM'd customers may be deprioritized (that's what I'd do if I had to STM people...) but then again if business customers paying double don't get prioritized I doubt they'll be doing that. It's also possible (and sometimes feels like) they run a complete free-for-all without per service-flow quotas or balancing, so if someone just happened to have more torrent seeds than you they may get more speed.

---------- Post added at 14:53 ---------- Previous post was at 14:50 ----------

Originally Posted by Kushan

I guess the only thing wrong with that is assuming there's an even distribution of bandwidth and that everyone can only get 20meg. I'm not really sure how the load balancing works, but if you're only getting 20meg, does that definitely mean everyone else is getting only 20meg, or are some getting more than that?
I really do wish we had more data on how effective STM is under heavy congestion.

When supposedly configured in a static, non-adaptive fashion across the whole country regardless of congestion levels it can't be very effective at all IMO. The only real effective way is the Comcast way and they seem to have deliberately abandoned that.

[Sirius]

Originally Posted by Kushan

we really need a better way of monitoring our data usage. Sadly, I get the feeling Virgin is putting this off as they're not monitoring it correctly themselves!

It cannot be that hard to introduce a means to monitor our own bandwidth usage

We used to be able to use snmp to monitor the modems but they blocked that pretty dam quickly

[qasdfdsaq]

Originally Posted by Sephiroth

My point, Kush, is that STM ion a saturated environment makes no speed difference to what the user would otherwise receive.

If they set everything up the lazy way, with no traffic prioritization, load balancing, or capacity management then yes, that's what would happen. But they have plenty of options at their disposal to change that, whether they use any of them I don't know.

Indeed it would be pointless because peak time on 8 channels would be peak time on the other 8 channels in a service group.

Not necessarily, that would require there be an equal distribution of user(s) and/or bandwidth demand across all channels, the same number of modems on all channels, and the same user demographic. It's easy enough to have a street full of businesses that shut up shop at 5pm, along the next road on the same node a pile of students, unemployed and other internet scrounging no-lifers, then an estate full of retirees, residential homes, and council housing bottom-feeders who can afford weed and gambling but no broadband.

Since you're implying VM aren't dynamically load balancing across all available capacity, then it follows that they aren't able to dynamically balance peak load either, and left to chance peak load is not likely to statistically normalize itself across all channels unless they're very very fat channels.

---------- Post added at 15:05 ---------- Previous post was at 15:04 ----------

Originally Posted by Sirius

It cannot be that hard to introduce a means to monitor our own bandwidth usage

We used to be able to use snmp to monitor the modems but they blocked that pretty dam quickly

The system is already there, they just have to create and link it into another system that allows you to see it.

[Sephiroth]

Originally Posted by qasdfdsaq

---------- Post added at 14:50 ---------- Previous post was at 14:47 ----------

[/COLOR]
Not necessarily. Again it'd depend on how well their load balancing is set up and the specific metrics used. STM'd customers may be deprioritized (that's what I'd do if I had to STM people...) but then again if business customers paying double don't get prioritized I doubt they'll be doing that. It's also possible (and sometimes feels like) they run a complete free-for-all without per service-flow quotas or balancing, so if someone just happened to have more torrent seeds than you they may get more speed.[COLOR="Silver"]

....

Ah yes. That's it. The ALLOT system deprioritises STM'd traffic. So congestion must be really bad if my un-STM'd speed tests frommy nearest VM host only manage 20 meg ofr much of the mid-evening.

[kwikbreaks]

Originally Posted by Sirius

We used to be able to use snmp to monitor the modems but they blocked that pretty dam quickly

Another reason to take the modem mode/own router option.

[Skie]

Originally Posted by Kushan

we really need a better way of monitoring our data usage. Sadly, I get the feeling Virgin is putting this off as they're not monitoring it correctly themselves!

Remember when a data usage monitoring widget appeared in the MY VM area? I bet one is 'coming soon', in the queue of all of the other things VM have coming 'soon'.

The heat death of the universe is soon, too.

[Kushan]

"soon" by VM's standards.

[qasdfdsaq]

Is about the same as "soon" in astronomical terms

Was it a widget for testers? I never saw it.

Yes 200mb was coming soon for the olympics after all..

[MUD_Wizard]

Originally Posted by Kushan

When was that? There was an issue a few months back but as far as I could tell it was largely resolved. It should now be measured every 15mins and at most last 2 hours (unless you continue downloading).

No it wasn't resolved. VM completely ignored our questions about 4 hour STM in that thread as I told you at the time. Now do you believe me?

The only thing they fixed at the time was a profile misconfiguration that caused some people in some areas to get STM'd at ~48Mb.


From our tests at the time, as far as I recall, we all triggered 4 hour STM by downloading as little as 6.1GB.

STM is measured every 15 mins, but can take a further 15 minutes to kick-in ; probably because one system (monitoring) has to communicate to another (limiting) and both are on a 15 minute timer. I found something about that at the time in some Docsis documentation on STM.

Anyway I've been on no STM since just after the Cat C Seg so haven't performed any STM tests since then. No point.