Terrible performance Leicester (LE3)

[Chrysalis]

no its just congestion relief, they are keeping shut tight on the upload upgrades.

[philce]

Originally Posted by Chrysalis

no its just congestion relief, they are keeping shut tight on the upload upgrades.

Why not just do the upgrades then? Makes sense otherwise 2 lots of work instead of one bigger one?

WTF do Hinckley seem to get everything before us? (FTTC now the VM upgrades?) Or is that not a coincidence!!!

[Chrysalis]

Originally Posted by philce

Why not just do the upgrades then? Makes sense otherwise 2 lots of work instead of one bigger one?

WTF do Hinckley seem to get everything before us? (FTTC now the VM upgrades?) Or is that not a coincidence!!!

Just the upgrades I expect wouldnt relieve the congestion. Or it could be the upgrades arent planned until late 2011.

[Ignitionnet]

Originally Posted by Chrysalis

no its just congestion relief, they are keeping shut tight on the upload upgrades.

It may not be the tier uplift but at that time scale, conveniently at the end of the calendar quarter and 4 months away which is a long time for resegmentation, it's extremely likely that'll be the pre-uplift work.

I'll poke someone and see what I can find out.

[uno]

My parents are still having lots of problems with slow speeds in the LE4 3 postcode area and have spoken to a couple of people in their street and report same problems so Leicester as usual really seems to struggle and last tech i asked said the upload increase would not be until summer 2011.

It does seem that Virgin are hugely overselling the network or have fell behind badly with network investment in the city also everybody i speak to says also keep getting a VOD error saying area is too busy so i presume again lack of capacity

[philce]

Originally Posted by uno

My parents are still having lots of problems with slow speeds in the LE4 3 postcode area and have spoken to a couple of people in their street and report same problems so Leicester as usual really seems to struggle and last tech i asked said the upload increase would not be until summer 2011.

It does seem that Virgin are hugely overselling the network or have fell behind badly with network investment in the city also everybody i speak to says also keep getting a VOD error saying area is too busy so i presume again lack of capacity

Its because BT refuse/cant/wont upgrade to FTTC and now we have the option of stable and fast broadband obviously we have all gone to Virgin.

They obviously take the money and until enough noise is generated by disgruntled users will do nothing.

I suggest you and your neighbours complain to the CEO of Virgin, that way the people who can do something about this will know about it. I would ring faults to check that it isnt a modem or cab fault first though.

[uno]

My parents have had a tech out as have the neighbours he changed a couple of bits including the modem but then went back to normal even he admitted when pushed that the network was not as good as it should be at present.

[Chrysalis]

Originally Posted by uno

My parents have had a tech out as have the neighbours he changed a couple of bits including the modem but then went back to normal even he admitted when pushed that the network was not as good as it should be at present.

I get the same story of local tech's they usually say things along the line of "area jam packer" "many customers in this area" "nothing I can do we ran out of capacity".

I can also confirm many people round here in same boat as me, I still got some work to do before I can get back to the newspaper tho.

[Chrysalis]

ok I would appreciate it if ignition or anyone with the answers can help here.

Currently we know LE3 has just 1 UBR.

Someone suggested on VM's forums that in busy areas UBR's would host 1000-2000 users, to me that sounds wrong. It would equal a 1.2% takeup for 2k subscribers. The takeup is over 60%.

So if every card slot was filled up in the UBR, and VM didnt oversubscribe what would be the max subscriber capacity?

[Ignitionnet]

LE3 doesn't have just one uBR it's covered by at least 2, one legacy and one overlay network.

The legacy CMTS, there will be more than one of these feeding LE3, is good for 2,400 connections depending on loading levels, 300 customers per 38Mbps downstream seems reasonable.

The Cisco 10k is good for tens of thousands. It has 8 line card slots each of which can support 20 downstreams and 20 upstreams, or it can feed an external modulator.

There are 15 CMTS in Leicester hubsite, another 13 in Northfields, LE7, at least 2 of which in each site are 10k or BSR.

I'm confused by the statement about 2000 users being a 1.2% take up though.

LE3 LEICESTER Braunstone, Glenfield, New Parks, Groby Road (A50), Leicester Forest East, Westcotes

I doubt that that area is a population of 200,000 given that the city's population is listed as 300,000 and remembering that a 'user' for cable purposes is a cable modem, and there's only one allowed per property. Even at 100% take up there would not be 200,000 cable modems in the whole of Leicester, the average home having over 2 residents per premise.

Braunstone apparently has about 7,500 households, Glenfield 5,000, New Parks 8,000, Westcotes 4,000, unsure about the other two and can't be bothered to search that hard but nonetheless it's not 2,000 households, 26,000 + those two areas so at 60% takeup a next generation CMTS plus a couple of legacy ones would be fine!

Your issues stem from a lack of 16QAM upstreams and the DOCSIS 2 upgrades, once all that's done along with the ongoing resegmentation that's being done for downstream and network quality purposes your upstream capacity issues will be gone with no need for any further additional upstream ports.

He says optimistically.

TLDR - the amount of chassis are fine.

[Chrysalis]

thanks for the reply, but whats this ongoing resegmentation? seems theoretical

I read on the VM forums that reseg's arent always upgrades but sometimes are just moving users around between existing line cards?

so 300 10mbit users on 38mbit? is about 19:1 contention right? if assume 1 in 5 users is a 20mbit customer it then becomes a bit higher tho around 24:1. That contention ratio seems reasonable but it is a very small fragile amount of shared bandwidth, evident by the downstream congestion on the legacy network.

Also you missing various areas under LE3, eg. I dont live in any of those areas. But thanks for making the effort to get the population count of the ones you found, I will do a bit more checking into it.

[Ignitionnet]

Originally Posted by Chrysalis

thanks for the reply, but whats this ongoing resegmentation? seems theoretical

I read on the VM forums that reseg's arent always upgrades but sometimes are just moving users around between existing line cards?

Yes, they are sometimes balancing load, say where there are 3 nodes on one card and a single node on another they will move a node to another card. This is an upgrade to the 3 nodes, all of them have more bandwidth per modem post-resegmentation.

The resegmentation programme is ongoing throughout that 'general area'.

---------- Post added at 12:02 ---------- Previous post was at 12:00 ----------

Originally Posted by Chrysalis

so 300 10mbit users on 38mbit? is about 19:1 contention right? if assume 1 in 5 users is a 20mbit customer it then becomes a bit higher tho around 24:1. That contention ratio seems reasonable but it is a very small fragile amount of shared bandwidth, evident by the downstream congestion on the legacy network.

Due to the relatively small customer population it generally works fine. Most of the 20Mb customers will be on the overlay network.

There's relatively little downstream congestion on the legacy network, it's largely overlay network upstream congestion that's the current bugbear.

Incidentally 300 x 10Mbit customers on 38Mbit is about 80:1 contention ratio

[Chrysalis]

ok but we know in le3 there is downstream congestion on legacy and even some on overlay albeit much less severe. its on at least 2 legacy ports as philce had it on legacy and I also had it on legacy. I expect many light takeup areas elsewhere in the country have much less than 300 per downstream as insufficent sales.

80:1 contention ratio is much higher than the amount I seen you post in another post, cant remember which site it was (think was tbb) and is quite a high number. Even on a large backhaul (eg. 1gigabit) that is high. We will disagree on if we think its reasonable, I have evidence to say it isnt in the performance we see here. Some VM areas have less than 50 on a downstream port.

So if resegmenting is sometimes only moving from one card to another (I assume the cards are in close proximity to each other) then why does it take many months to do the work? and what happens if all cards are saturated?

[Ignitionnet]

Originally Posted by Chrysalis

ok but we know in le3 there is downstream congestion on legacy and even some on overlay albeit much less severe. its on at least 2 legacy ports as philce had it on legacy and I also had it on legacy. I expect many light takeup areas elsewhere in the country have much less than 300 per downstream as insufficent sales.

You expect wrong If there are light sales what's the point in spending money on hardware and resegmentation to leave all that capacity unused?

At one time 200+ modems per upstream wasn't uncommon when the speeds were lower, 5,000 cable modems on a single chassis not impossible. Where areas have light utilisation VM simply wouldn't have split them as much as heavier areas, so same kinda number of modems online just more homes passed per card.

Your area is the exception rather than the rule downstream, as you will have noted from the forums the overwhelming majority of issues are upstream.

Originally Posted by Chrysalis

80:1 contention ratio is much higher than the amount I seen you post in another post, cant remember which site it was (think was tbb) and is quite a high number. Even on a large backhaul (eg. 1gigabit) that is high. We will disagree on if we think its reasonable, I have evidence to say it isnt in the performance we see here. Some VM areas have less than 50 on a downstream port.

The 30:1 I mentioned was for overlay network.

Your issues are upstream related I believe? No idea what the ratio in your area actually is. Probably somewhat higher than the number I mentioned.

I would like to see these areas with 50 modems on a downstream port, they would be few and far between indeed. 50 on an upstream port for sure, 50 on a downstream not so much. Certainly last year most ports that I saw were running between 300 and 450 modems, I wouldn't be surprised if that were 150 - 200 now in most cases.

Reasonable contention ratios are as high as companies can get away with, they have become largely meaningless due to the various bandwidth reducing methods used. In VM's case shaping and STM are the big ones that allow the network to run at higher ratios than would normally be feasible.

Regardless the levels of contention operators run their networks at are generally obscene

[Chrysalis]

I meant not upgrading to reduce to that level but rather that original sales werent enough to get higher usage.

why is legacy run at a different contention to the overlay network? thats an interesting one.

to me contention ratio is always relevent, its the ultimate measure of budgeted bandwidth. You said it yourself, the higher contention is only barely achievable due to STM and shaping been used to cut costs.

I agree on your last point.