VMBU showing erroneous data
Sorry for not responding quicker, got side-tracked elsewhere, and stomach was rumbling.
Hi Barry,
Thanks for that explanation, that at least restores confidence in the VMBU totals that show on the main page :), assuming that BT haven't got any exchange or other problems ???, I'll have to keep a careful eye on their clock
Makes perfect sense. So considering the way we are charged then that all seems perfectly reasonable and fair.
As the VMBU breakdown is only shown in %, it's a bit difficult to make any exact calculations. But as my free usage this month has been low so far, taking the 24% from the breakdown, of my VMBU main page total gave a figure some 26% different from the VMBU free total
so James's use of the word 'massive' would seem appropriate here 
.
Whilst I'm not going to get over-excited about that as the actual Mb values aren't huge, taking eg. 2% of say a 1GB monthly allowance, some 20MB, I would like to know what type of usage would get mistaken for that amount of gaming? I don't use any unusual protocols, mainly browsing, a bit of e-mail (insignificant this month so far), saving a few web pages and some pdf files, one file download so far (38MB), and a very odd bit of video streaming (usually off-peak).
@Phil Richardson, whilst I appreciate you are trying to be helpful, your very simple analogy may help those who couldn't follow Barry's explanation, but you don't address the primary issue -
If it's large enough to appear as a % on the breakdown, then it is of concern.
I've seen reference to the Ellacoyas elsewhere, in relation to VMBU but if I'm not mistaken, also for traffic shaping, but I've skimmed over a lot and not retained much, so I will have to do some more background reading before I can make further (constructive I hope) comment, but I have a feeling that as this issue has obviously been around for some time (why else is it mentioned in the support pages) then perhaps there should be a few red faces at PN towers for not getting this addressed, after all the average user would expect typical usage to be shown relatively accurately.
Hi Barry,
Quote from: Barry As I understand it, the VMBU totals are received directly from BT via their RADIUS logs.........................the BT totals sent through from RADIUS are the ones used to calculate whether you exceed your allowance or not, and I believe these are fairly applied.
Thanks for that explanation, that at least restores confidence in the VMBU totals that show on the main page :), assuming that BT haven't got any exchange or other problems ???, I'll have to keep a careful eye on their clock
Quote from: Barry ..............................there is potentially a chunk of usage that isn't recorded in the VMBU 'totals'. However, the Ellacoyas are likely to record at least some, if not all, of this usage so the VMBU breakdown will end up totalling more than the VMBU totals (if that makes sense).
Makes perfect sense. So considering the way we are charged then that all seems perfectly reasonable and fair.
As the VMBU breakdown is only shown in %, it's a bit difficult to make any exact calculations. But as my free usage this month has been low so far, taking the 24% from the breakdown, of my VMBU main page total gave a figure some 26% different from the VMBU free total
so James's use of the word 'massive' would seem appropriate here .Whilst I'm not going to get over-excited about that as the actual Mb values aren't huge, taking eg. 2% of say a 1GB monthly allowance, some 20MB, I would like to know what type of usage would get mistaken for that amount of gaming? I don't use any unusual protocols, mainly browsing, a bit of e-mail (insignificant this month so far), saving a few web pages and some pdf files, one file download so far (38MB), and a very odd bit of video streaming (usually off-peak).
@Phil Richardson, whilst I appreciate you are trying to be helpful, your very simple analogy may help those who couldn't follow Barry's explanation, but you don't address the primary issue -
Quote from: P ......................................................... breaking down how it is used is not so simple, and as per the statement on the support pages, some will appear in incorrect categories. It's when you are seeing more than you currently are that there should be a concern.
If it's large enough to appear as a % on the breakdown, then it is of concern.
I've seen reference to the Ellacoyas elsewhere, in relation to VMBU but if I'm not mistaken, also for traffic shaping, but I've skimmed over a lot and not retained much, so I will have to do some more background reading before I can make further (constructive I hope) comment, but I have a feeling that as this issue has obviously been around for some time (why else is it mentioned in the support pages) then perhaps there should be a few red faces at PN towers for not getting this addressed, after all the average user would expect typical usage to be shown relatively accurately.
Quote from: Anotherone I've seen reference to the Ellacoyas elsewhere, in relation to VMBU but if I'm not mistaken, also for traffic shaping, but I've skimmed over a lot and not retained much,
The Ellacoyas are technically 'accounting' servers. Their main function is to analyze a packet of data (or a datastream) and identify what sort of traffic it is.
Once they have identified some data, there are several avenues open to them.
a) Record the type of traffic against the customer, to display the VMBU data (the true 'accounting' purpose)
b) Set the QoS (Quality of Service) field in the packet so that all of the Plusnet hardware knows how to prioritise that particular packet across the network
The actual traffic shaping isn't done by the Ellacoyas, I don't believe. That job is handled by another part of the network (I would hazard at guess at the routers that connect the centrals to the network). However, they do work on the premise that the Ellacoyas identify the traffic first.
Prioritising the traffic is actually more complicated than it seems. Of course, it's based on identifying the packet in the first place, but that information then has to be combined with the user and their particular product (i.e. they're on BBYW Pro). They also need to check whether that user is restricted (i.e. they've used up all their included bandwidth). It also has to be combined with the various hourly schedules (i.e peak time), and also the network condition (Plusnet has the ability to divert 'spare' network capacity to allow protocols that are usually low-priority to soak up any leftover allocation after all the gold/platinum traffic has been accounted for). There is also an 'overall' network status which can be brought into effect (for example if they have a central drop) which can lower the priority of some traffic on an 'emergency' basis.
So, all of this is calculated in real-time as the packets flow across the Ellacoyas.
As to why you're getting a 'significant' amount of traffic identified as Gaming - I suspect it's straight-off misidentification on behalf of the Ellacoyas. it may even be 'Internet Background Noise' such as port scans - some of which may be identified as Gaming purely because they scan *all* available ports.
Obviously, there are gaps in the ruleset that feed the Ellacoyas. Dave Tomlinson posted here a brief overview of how rules get added. The important thing to bear in mind that the more rules that the Ellacoyas have to process, the longer it will take (or the more likely that the resource limits on the devices themselves will start being a problem). Whilst they're operating in sub-millisecond times at the moment, it would cause massive problems if they started inducing lag on the very traffic they're trying to prioritise as titanium!
Anyhow, I think I meandered around the topic enough already, so I'll stop there and wait for Tommo to come and correct where I'm wrong
B.
EDIT: I forgot to add that it's probably likely that the Ellacoya's don't work on a strictly 'store-and-forward' mechanism. I wouldn't be surprised if they let the first few packets of data through either unidentified, or prioritised until the particular stream has been identified and placed into the correct prioritisation queue.
@ChrisL
You've hit the nail on the head.
@Barry Z
Thanks for a very clear and concise insight into the job of the Ellacoyas. It's clarified what was a fuzzy recollection at the back of my brain somewhere, not having yet had the chance to do that background reading. The link was also useful.
I shall add your post to my favourites for future reference (sorry should I say 'Bookmark' ).
The only bit I couldn't really get to grips with was the bit about why traffic may be mistakenly identified as Gaming
I don't use any port scanning software (that I know of!), my firewall is set to only allow specific programs access (IE, OE, AVG update etc), everything else has to ask. So are you saying that in effect (DS) port scanning traffic, even if it doesn't get past the PN firewall, gets counted as usage?
You've hit the nail on the head.
@Barry Z
Thanks for a very clear and concise insight into the job of the Ellacoyas. It's clarified what was a fuzzy recollection at the back of my brain somewhere, not having yet had the chance to do that background reading. The link was also useful.
I shall add your post to my favourites for future reference (sorry should I say 'Bookmark'
).The only bit I couldn't really get to grips with was the bit about why traffic may be mistakenly identified as Gaming
Quote from: Barry ................... it may even be 'Internet Background Noise' such as port scans - some of which may be identified as Gaming purely because they scan *all* available ports.
I don't use any port scanning software (that I know of!), my firewall is set to only allow specific programs access (IE, OE, AVG update etc), everything else has to ask. So are you saying that in effect (DS) port scanning traffic, even if it doesn't get past the PN firewall, gets counted as usage?
