Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Chrysalis]

plusnet (and most other isps) would be congested way before the cabinet is plus the fibre links can do more than a gigabit of throughput.  Of course on top of that many cabinets are not full and not everyone will be synced at 80mbit, most dont even sign up to the 80/20 package.

also hopping doesnt just sort it briefly, I have been ok for a week or so now, and so has kitz it seems.

So if we assume say avg 40mbit per customer, and every customer is a torrent nutcase, it still would take 200 customers to reach 800mbit of throughput, this is an extremely low contention ratio compared to the isp's backhauls.

[tommy45]

Well since the other night  i to may of landed on a good endpoint , but only time will tell,  no reduction in throughput at peak times or jitter on the ping monitor , so far but although ,maybe not related  it have noticed packetloss to Steampowered.com  from the plusnet hop immediately before level3 to end of destination  again during the same times as plusnets peak time, 

[boost]

plusnet (and most other isps) would be congested way before the cabinet

I find that difficult to believe. These cabs are almost certainly provisioned on an over-subscription model and there are almost certainly some use cases where it doesn't align with The Spreadsheet.

also hopping doesnt just sort it briefly, I have been ok for a week or so now, and so has kitz it seems.

Sounds like they've isolated the fault then? Excellent!

So if we assume say avg 40mbit per customer, and every customer is a torrent nutcase, it still would take 200 customers to reach 800mbit of throughput, this is an extremely low contention ratio compared to the isp's backhauls.

I mentioned torrents in the context of them being able to ravage processor usage. The bandwidth they consume is just an aside!

Well since the other night  i to may of landed on a good endpoint , but only time will tell,  no reduction in throughput at peak times or jitter on the ping monitor , so far but although ,maybe not related  it have noticed packetloss to Steampowered.com  from the plusnet hop immediately before level3 to end of destination  again during the same times as plusnets peak time,

Control plane policing is often the reason for the erroneous packet loss you see in the middle. Big expensive looking routers don't care about traffic directed AT them, they only care about the traffic going THROUGH them

I'm just e-clutching at straws though tbh. I'm sure there must be someone on here with intimate knowledge of all this kit...

[ejs]

So if we assume say avg 40mbit per customer, and every customer is a torrent nutcase, it still would take 200 customers to reach 800mbit of throughput, this is an extremely low contention ratio compared to the isp's backhauls.

Sorry, but shouldn't that be 20 customers?

[Chrysalis]

you right it isnt 1:1 contention but its low enough that you will see 1:1 performance.

I believe tho the upload speeds are uncontended.  That is how highly provisioned they are, remember its expensive for openreach to deploy fibre, so they have done excess amounts of it as they wont want to do it again.

In short the bandwidth provisioned per end user at a cabinet level is far higher than what the isp's provision.

I think the most realistic point of congestion nearest the customer is the SVLAN and mine is confirmed BLUE.

Of course as well as you pointed out, there is no way 200 users are going to be maxing their lines out at the same time.

Sorry for my bad maths it would take 20 40mbit customers to reach 800mbit of throughput. A 10:1 contention ratio (or there abouts) for a consumer broadband product is very low.  The isp's will be contending way higher then that.

Also a quick read of the plusnet forums, doesnt suggest the problem is fixed, merely just me and kitz have found good endpoints and kept our PPP sessions running since.

Also i forgot to mention, there is a openreach document somewhere that states how much they provision per customer, if i remember correctly its basically 4:1 contention ratio as a worse case scenario (20mbit provisioned for 80mbit customers) meaning if 200 users are in the cab synced at 80mbit, then there will for sure be more than 1 gigabit port connected.

[tommy45]

Well the usual jitter is back tonight , and i'm seeing some strange  latency effects to the BBC  the latency as from earlier today has also increased to the bbc by 8-10ms

Code: [Select]

|------------------------------------------------------------------------------------------|
|                                      WinMTR statistics                                   |
|                       Host              -   %  | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
|               home.gateway.home.gateway -    0 |  101 |  101 |    7 |   14 |   23 |   13 |
|     lo0.10.Central10.ptw-bng02.plus.net -    0 |  100 |  100 |   18 |   33 |   89 |   31 |
|                irb.10.PTW-CR01.plus.net -    0 |  100 |  100 |   18 |   29 |   63 |   24 |
|              kingston-gw.thdo.bbc.co.uk -    0 |  100 |  100 |   18 |   31 |   67 |   24 |
|                   No response from host -  100 |   19 |    0 |    0 |    0 |    0 |    0 |
|                   No response from host -  100 |   19 |    0 |    0 |    0 |    0 |    0 |
|                ae0.er01.telhc.bbc.co.uk -    0 |  100 |  100 |   19 |   31 |   48 |   30 |
|                         132.185.255.149 -    0 |  100 |  100 |   19 |   30 |   84 |   26 |
|               fmt-vip71.telhc.bbc.co.uk -    0 |  100 |  100 |   19 |   29 |   73 |   28 |
|________________________________________________|______|______|______|______|______|______|
   WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider
Were as when i run the same test to

Code: [Select]

|------------------------------------------------------------------------------------------|
|                                      WinMTR statistics                                   |
|                       Host              -   %  | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
|               home.gateway.home.gateway -    0 |  104 |  104 |    0 |    0 |    1 |    0 |
|     lo0.10.Central10.ptw-bng02.plus.net -    0 |  104 |  104 |   11 |   13 |   51 |   11 |
|                irb.10.PTW-CR02.plus.net -    0 |  104 |  104 |   11 |   12 |   29 |   11 |
|                 lonap-gw1.thdo.ncuk.net -    0 |  104 |  104 |   11 |   16 |  203 |   12 |
|       te1-1-1-31.core-rs2.thdo.ncuk.net -    0 |  104 |  104 |   12 |   15 |   51 |   12 |
|             pingbox1.thinkbroadband.com -    0 |  104 |  104 |   11 |   12 |   22 |   12 |
|________________________________________________|______|______|______|______|______|______|
   WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider   Gives normal results (apart from the variation in latency/jitter)
Maybe down to plusnets perfect networking that's never at fault ,apparently

[Chrysalis]

iplayer started buffering on my amazon firetv box so I hopped onto my pc to see some disturbing speedtests.

Both download and upload erratic so about to do a hop, I did 5 tests in a row to see if results consistent and they are.

post hop



pre hop




iplayer not retested yet tho.

Although the average speeds were still high it was very spiky on dumeter, and a dodgy upload meaning something clearly wasnt right.

[Chrysalis]

iplayer is fine again.

some more info.

After the hop I am back on the same plusnet gateway, although I could well be on a different endpoint.

The iplayer buffering was very extreme, basically it would be a spinning circle for over a minute and then play just one second.

[Chrysalis]

to kitz, tommy and co, seems jelv did some digging.

http://community.plus.net/forum/index.php/topic,136571.msg1198558.html#msg1198558

I've just searched the service status archive - the last announcement of increased capacity was last July (June for 20CN which I am on).

My gut feeling is that we are back in the bad old days of Plusnet not having enough capacity. The rate at which they've been installing FTTC connections we should have been seeing regular announcements

[kitz]

Thanks for pointing out that post.    I am getting a weird felling of deja vu over there.

I cant really say much atm though since Im not getting home until late evenings, but afaik everything has been fine since my last pipe hop and its why Im staying on this pipe.   Before I got on this pipe it was bad for weeks.

[Chrysalis]

also here is the tbb graph, note the before and after red line in evening.

[kitz]

Mines here.

btw Ive noticed 2hrly peaks on my graph for a while and note that there also appears to be some evidence of it on yours.   Im pretty certain its not coming from this end.   I keep meaning to isolate the PC and turn off wi-fi one day, but that means it would also turn off dslstats and hg612 modem stats.

[NewtronStar]

, but afaik everything has been fine since my last pipe hop and its why Im staying on this pipe.   Before I got on this pipe it was bad for weeks.

Seemed to be stuck on one pipe for few months on BTw (Carlisle) even after numerous new PPP sessions 

[tommy45]

Well 8 days is the record for me in the past 2 and a bit months , again tonight  i get plusnot's jitter and poor throughput  hopping gateways took me  an age to find another gateway or/and endpoint that was able to provide a slightly better throughput, there was several gateways /endpoints that where providing less throughput , they typically also had higher base  latency and jitter levels  hopping gateways is not real answer  it's like a minefield,  Its also about time plusnet owners BT started buying more capacity  instead of investing further into the race to the bottom

cash backs free months subs, bb & phone & free tv deals,  need to go

[Alucidnation]

I have to say I'm beginning to regret joining Plusnet.

Web pages stalling etc.