Print

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

[boost]

you work for an isp to be able to see that?

Of course. A small no-name with no power
I get very few FTTC faults so it's been a long time since I needed to look at that screen but I'm pretty sure the cvlan stuff is newish

[Chrysalis]

ok if you ever come across one without 0s then you will have my attention

As far as I know each cabinet only has one CVLAN, so basically its a CVLAN per cabinet, possibly multiple cabinets per CVLAN.  But not more than one CVLAN per cabinet.  Which should mean it would be impossible to change CVLAN's without physical presence.

[boost]

You may be right but that doesn't make any sense in my mind.

We already know that customer traffic appears to be running on VLAN 101 with BT traffic running on VLAN 301. The HG612 shows that PTM 301 traffic runs at COS 5 (prioritised) and 101 runs at 1~. You have to have a VLAN tag in order to write the 802.1p bits so this suggests to me the entire frame makes it to the MSAN. The MSAN probably strips it and adds a customer specific one but even at this early point, there are many things which could be considered the CVLAN.

I might well be missing something / oversimplifying though?

[guest]

I see Andrew over at TBB has started a thread entitled "Are PlusNet speeds in meltdown?"

That implies (to me) he's seeing speedtest stats which suggest there is something amiss with Plusnet & only Plusnet and is looking for confirmation before an article.....

[tommy45]

I see Andrew over at TBB has started a thread entitled "Are PlusNet speeds in meltdown?"

That implies (to me) he's seeing speedtest stats which suggest there is something amiss with Plusnet & only Plusnet and is looking for confirmation before an article.....

Asking around other isp's i would say that the bulk of the problem has been and what remains is  unique to Plusnet ,As some of the earlier reports of issues back in December 2014  where down to BTW  network limited to a few areas, was apparently resolved by in Jan/FEB 15 , these where well documents by other ISP's including AAISP
As well increased Jitter and base latency levels during peak times,packet loss has been seen too, and some external peering  shows a bigger increase in latency (on some links)

But as a customer for nearly 2 yrs  i can categorically say that they have more often than not had these same type of issues (except for packet loss) which was  mostly confined to a Sunday or and Monday evening's, but as of late this has extended into an almost nightly event , though some nights are not as bad as other nights, frequently having to drop PPPOE session  in order to  connect to a less saturated  endpoint ,This also of late has been requiring a lot more attempts So is becoming troublesome and frustrating  before achieving  the desired effect,

And when these issues are going on  the like of I player or ITV player are also impacted on   streams are interrupted (buffering )

As for the thread  you referred to i think that is dead alreadyhttp://forums.thinkbroadband.com/plusnet/t/4398100-conclusion.html If that is his final word on it

[kitz]

Its my understanding that the CVLANs are between the DSLAM and the OLT.   Ive seen a doc somewhere stating that the CVLANS are fixed and cant be changed as eventually they form an 'inner part' of the SVLAN.  ie many CVLANs in the one SVLAN.  The C for CVLAN stands for customer and is set when the line goes live and the customer retains that same CVLAN for the service lifetime. 


In the case of GEA fibre, the SPs (such as sky) are handed over the CVLAN which is customer specific and never changes. I'm uncertain when they say customer specific whether they mean all EU's from that DSLAM to the OLT will have the same CVLAN and if a particular DSLAM can have more than one CVLAN. 

Requests can be made though for a change of ServiceVLAN which is what does the QoS and has a limited amount of bandwidth.    Many SVLANs can be within in the one cable... ie these are 'partitioned' parts of the total cable bandwidth and differently tagged SVLANs may get priority over others.  My understanding is that any priorities ie 100/301 tagging are at the SVLAN level.

Edit this could imply the CVLAN is line specific - not DSLAM

A Customer VLAN
(CVLAN) is created when a service (e.g. Data or Voice) is provisioned for one of your customers. At this
point you may specify a particular Service VLAN (SVLAN) or no SVLAN in which the CVLAN is placed. You
can create new SVLANs by raising a Modify order against your GEA Cablelink product

[kitz]

Ive just had a quick look over there.   Not sure there is much point posting because someone seems to take it on their behalf to reply on behalf of Plusnet.

What Andy says about the host links is true and can easily be confirmed on the BTw website.

However I agree for me there's been an awful lot of deja vu over the past six months since the drops on the BNGs late last year.   Some of you may recall at the time that it was like a repeat of history when the Junipers way back in about 2005 couldnt cope with the number of sessions and I put that forward as a suggestion.  Then I later see that theyve had to limit the MX960's to 50k users 

They also denied the first time round about capacity issues, but it could be seen way more clearly - ie that the graphs were getting full and flat-lined.    The graph on Sat night showed indication of this, but it does usually seem to be ok.

 I really cant understand why the Ellacoyas arent properly controlling latency and jitter such as jelv is seeing.   This should not be happening if the issue was on PN's network.
... but back to the fact that you cant now buy ellacoyas and its Proceras.    I wonder if they are having teething problems getting config right like they originally did with the ellacoyas.      Im only throwing suggestions because deja vu was mentioned..  but if it was that then you'd only expect the issue to happen on the BNGs which is where I assume theyve purchased proceras for.

However, you cannot exceed the physical capacity between the WBMC handover nodes and your network (irrespective of any burst caps) without traffic being dropped.

This I cannot understand...   If it were an issue then PN should be using the Ellacoyas/Proceras to ensure that jitter doesnt occur and no packets are dropped.   It just doesnt make sense from here what is going on. 

[jelv]

I'm on 20CN via IPSC which never connects to a bng!

They did suggest it might be something to do with being on the IPv6 trail so for a week or so I reverted to my normal connection (Extra) with IPv6 disabled on router and PC. I was still getting issues with VoIP call quality and Lync losing connection during the evening.

[boost]

Install MX960s they said. You'll be able to inspect at line rate they said. Don't worry about bugs they said

Far easier to let the community to squabble over perceived bandwidth problems whilst you write the press release about how your spying efforts are going horribly awry.

jelv, how do you know you never connect to a bng? You could be tunnelled through any one of them. Isn't that the whole point of IPSC?

I think you have it right, kitz. The nomenclature is very confusing but assuming metro speak, then the L2S needs to be capable of receiving untagged/single or double tagged from the MSAN (multi-VRF CPE attached to modem) all of which I would assume fall under CVLAN in BT spake.

Isn't this fun?

[kitz]

how do you know you never connect to a bng?

The IPSC connections are directed to specific gateways: ie the older AGs (E320's and ERX's).
The newer BNGs are those with the MX960's and 10Gb host links.

It shall be interesting to see how my line goes now -  Due to a resync caused by the dslam/exchange equipment,  Ive lost that nice gateway I was on.  Im now on a totally different gateway and at a different colo.

lo0.12.central12.pcl-bng02.plus.net [195.166.130.153]  (old)
lo0.11.Central11.ptn-bng02.plus.net [195.166.130.191] (new)

[Chrysalis]

Well I need to take a break from the plusnet forum, had enough of it.

[kitz]

Chrys.   I noticed on there that you said you had issues with the upstream.   

That should never be a problem for the host links which are fully symmetrical.   There should always be plenty on the upstream.   

If you look at their graphs upstream doesnt ever reach anywhere near the limits.   It never has done and its why some ISPs never included upstream for bandwidth caps. Its one of the reason why when I see someone's upstream graph doing the wavy lines or doing a slow start, I tend to discount contention... and its pointing to something else. 

[tommy45]

I have notice that the upstream will almost pause, drops to sub 1mbps sometimes then back to full again , i have only seen this happen around the peak time on the tbb speedtest ,as it appears to only affect single threaded, In testing i ran a single threaded file upload using FTP  and it wouldn't stay anywhere near the max throughput, it was yo yo ing , but at a quieter time it will upload at full tilt , so ? something is a bottleneck but for me this isn't really new, as i have been seeing similar since shortly after my service went live ,on and off for around 18mths , mainly sun & mon peak time

I read one of the REV k's rants and some of his customers where having issues , he put them on elevated traffic, and there was an improvement , yet btw were saying not faults ect iirc, if plusnet would for testing  switch off their traffic shaping  for a week or so, if it still persists  then they could put a number of those seeing issues on elevated  to see if that showed improvements or not, process of elimination maybe is what is required,  or some truths from plusnet, because whatever is /are the reasons for it it has been allowed to go on for too long already, plusnet saying we don't know  or it isn't our fault, ain't good enough

[Chrysalis]

Chrys.   I noticed on there that you said you had issues with the upstream.   

That should never be a problem for the host links which are fully symmetrical.   There should always be plenty on the upstream.   

If you look at their graphs upstream doesnt ever reach anywhere near the limits.   It never has done and its why some ISPs never included upstream for bandwidth caps. Its one of the reason why when I see someone's upstream graph doing the wavy lines or doing a slow start, I tend to discount contention... and its pointing to something else. 

yep which is why I said I suspect issues with something cpu intensive such as their ellocoyas.

[roseway]

Chrys.   I noticed on there that you said you had issues with the upstream.   

That should never be a problem for the host links which are fully symmetrical.   There should always be plenty on the upstream.   

If you look at their graphs upstream doesnt ever reach anywhere near the limits.   It never has done and its why some ISPs never included upstream for bandwidth caps. Its one of the reason why when I see someone's upstream graph doing the wavy lines or doing a slow start, I tend to discount contention... and its pointing to something else. 

That's interesting, because several days ago my upstream connection became so slow that the connection was virtually unusable. I did a gateway hop and everything was working normally again.