From what I understand from wombat's post, these rules are used to determine the "initial" capacity provisioned. It doesnt say anything about rigidly sticking to it and ignoring actual historical demand from customers. So e.g. if an area only performs well with say 5x tmax, then the 5xtmax should be preserved on a upgrade program, not reverted to say 2x tmax or consumption + tmax.

Whatever system, VM is using, the procedures they have in place for upgrades and so forth I see as inadequate, even if that means me saying the industry accepted practice is inadequate, but having read what wombat posted, that seems to just be a guideline for area's where there is no historical data so used as a "starting point".

There is also no mention in terms of usage limits, as a service selling limited usage will inevitably have lower demand put on the capacity.

Edited by Chrysalis (Tue 28-Feb-17 12:18:36)

Thanks for that link...

One common author between that doc and the two I had, which is Cloonan, the CTO of Arris.

Ironically, my documents are a year later than that one. Your doc presents Arris' discussion of future QoE methods, while my two appear to have settled on one known as "Network Quality of Experience" (NQoE), by Emmendorfer. They are actively bigging it up.

Anyway, this threadlet started relating to the assertion of MSO's who used a simple multiple ... and it seems clear it did happen. What it means for VM is different though.

For me, paying more attention in the DSL world, these documents throw up some different considerations - relating to the calculations on backhaul needs for PONs and G.Fast.Those, presumably, will operate in a manner similar to very small service groups.

Interesting to see, then, that Arris don't think there is much value in having service groups drop below around 250 premises.

I think you have to look into the details of the documents. Those suggest that the sizing rules have gradually shifted from being based on "historical demand" (or consumption) towards being based just on the tier size.

That suggests that node splits would originally happen when consumption got high enough, whereas the trend is towards node splits when marketing wants to sell a new top tier speed.

The caveat still remains - it is generic US information, so really hard to apply to VM.

then that is just bad logic. As we know e.g. DSL has long stopped using fixed contention ratio's as its not a viable system, user's bandwidth consumption is a constantly shifting metric.

I have made the assumption VM use this system based on ignition saying VM follow industry standards.

Your comment about node splits been dictated by marketing however does match what my own experience was when a VM customer, in that the only time I know for sure my area had a node split was when VM started their 10:1 products. Up until then VM seemed to just refuse to carry out the work.

My area I think has now had some more re segmentation as I think fibre has been pushed much deeper based on some work I seen during the new year period, but thats just me guessing on whats been done, as I seen them feeding some new cables.

Edited by Chrysalis (Thu 02-Mar-17 17:01:56)