Capharnaum
join:2006-06-19
Montreal, QC
| reply to DSL_Ricer
Re: The Bell Disclosure!
said by DSL_Ricer  :If you were trying to call 911 on a voip phone during that 0.37% of the time, would you be OK with the congestion? Congestion doesn't mean your VOIP 911 call wouldn't go through. It means that congested dslams get to "critical" levels, but it doesn't mean that it is overloaded and wouldn't process your call. Also, it is unclear whether congestion would just slow down the links or whether it would lengthen response time and by how much.
There isn't any data that lets us think the "congestion" would affect anyone for any particular length of time. |
|
Daryl Lamonica
@teksavvy.com
| reply to DSL_Ricer
said by DSL_Ricer :Now can someone please explain to me exactly from where to where in the network diagram their DSLAM category is, and why they can't selectively target traffic on congested links? I'm not an expert in Bell's network, so if there are any mistakes in this explanation they are mine:
In simplistic terms, a TS customer's data will traverse the following path:
a) the copper from your home to the CO (DSLAM) is speed restricted by the card in the DSLAM with provisions your service, and by the wire distance from your house to the CO. Currently Bell's DSLAM cards max out at about 7Mpbs as short distances from the CO. Typical data rates are 3-6Mbps downstream, depending on the service you purchased and the distance from the CO to your home.
2) From the DSLAM the signal goes to a concentrator (I'm using this terminology because the type of equipment used may vary) which then puts your traffic and that of other people onto a GigE type circuit (typically).
3) From there the data leaves the CO and heads to a NAP (like 151 Front St. W. in Toronto).
4) At the NAP, the data is passed over from Bell to TS's own co-located network and then onto the connections TS purchases from other providers.
I'm not 100% certain where Bell's Ellacoya boxes are physically located, but logically they could be located at either a step 2a or step 3a, and possibly even both locations to deal with upstream/downstream traffic before they hit the GigE pipes in either direction. |
|
TakeTheFifth
join:2004-04-20
Anjou, QC
1 edit | reply to DSL_Ricer
said by DSL_Ricer :Now can someone please explain to me exactly from where to where in the network diagram their DSLAM category is, and why they can't selectively target traffic on congested links? Because these would be the links carrying their Optimax traffic ?
But seriously, do these numbers include ADSL2 dslams ?
Phil |
|
mr_hexen
join:2007-08-02
Brampton, ON
1 edit | reply to DSL_Ricer
said by DSL_Ricer :said by mr_hexen :I did the numbers on page 1 or 2. it's 0.37% time. if the available time over 14 days is congested as little as 0.37% they mark the entire 14 days as congested 24/7. riiiight. If you were trying to call 911 on a voip phone during that 0.37% of the time, would you be OK with the congestion? A couple month ago Teksavvy failed to account for how long it would take Bell to install another Gig-E. That was a pretty awful few weeks. Bell's criteria for considering a link congested isn't all that bad. Their decision to solve it with DPI, however, was. Now can someone please explain to me exactly from where to where in the network diagram their DSLAM category is, and why they can't selectively target traffic on congested links? I dont have VOIP, and dont want it simply because I don't trust it. IMO, VOiP is too unreliable (whether due to network issues or not) to become the SOLE link for emergency services (as proved by that bc lady who moved and it all got screwed up when she called 911).
the DSLAM is the CO. This is where the 2 wires dedicated to your house get lumped into a big connection back to their offices in Toronto (the BAS, Broadband Access Server). Simply put, if a DSLAM is congested its because they oversold it. 1 port, 1 user. PERIOD. |
|
DSL_Ricer
Premium
join:2007-07-22
| reply to mr_hexen
said by mr_hexen :I did the numbers on page 1 or 2. it's 0.37% time. if the available time over 14 days is congested as little as 0.37% they mark the entire 14 days as congested 24/7. riiiight. If you were trying to call 911 on a voip phone during that 0.37% of the time, would you be OK with the congestion?
A couple month ago Teksavvy failed to account for how long it would take Bell to install another Gig-E. That was a pretty awful few weeks.
Bell's criteria for considering a link congested isn't all that bad. Their decision to solve it with DPI, however, was.
Now can someone please explain to me exactly from where to where in the network diagram their DSLAM category is, and why they can't selectively target traffic on congested links? |
|
mr_hexen
join:2007-08-02
Brampton, ON
| reply to Capharnaum
said by Capharnaum :said by TakeTheFifth :What I'd like to know is how/where they measure congestion at the dslam. 5.2% of the links are congested. Which ones ? Phil Remember that criteria for a congestion isn't that there's an overload. It's just that their "limit" (which isn't 100%) was passed four times out of 1344 checks within a two week span (if I understand the documents correctly). Then that dslam is labeled as "congestioned". It doesn't mean that the dslam was overloaded at any time during the month. It's pretty shady. I did the numbers on page 1 or 2. it's 0.37% time.
if the available time over 14 days is congested as little as 0.37% they mark the entire 14 days as congested 24/7.
riiiight. |
|
Capharnaum
join:2006-06-19
Montreal, QC
| reply to TakeTheFifth
said by TakeTheFifth :What I'd like to know is how/where they measure congestion at the dslam. 5.2% of the links are congested. Which ones ? Phil Remember that criteria for a congestion isn't that there's an overload. It's just that their "limit" (which isn't 100%) was passed four times out of 1344 checks within a two week span (if I understand the documents correctly). Then that dslam is labeled as "congestioned". It doesn't mean that the dslam was overloaded at any time during the month.
It's pretty shady. |
|
TakeTheFifth
join:2004-04-20
Anjou, QC
| reply to R0CKY
said by R0CKY :Definitely an interesting table... no doubt! What I'd like to know is how/where they measure congestion at the dslam. 5.2% of the links are congested. Which ones ?
Phil |
|
Kareeser
hm?
Premium
join:2006-07-18
Hamilton, ON
 ·Bell Sympatico
·TekSavvy Solutions..
| reply to yabos
I'm not sure whether 2-5% congestion means "The pipes are at an average load og 2-5%", or whether that means "2-5% of our pipes are fully utilized, and people are oversubscribed in those areas and getting substandard service"
I'm leaning more towards the latter, and 2-5% might encompass several neighbourhoods in Downtown T.O.! |
|
CitiLaptop
@teksavvy.com
| reply to yabos
Their definitions of congestion are pretty low too.
They deem it "congested" if the following utilizations are exceeded:
DS-3: 61%;
OC-3: 84%; and
OC-12 and OC-48: 90%.
And they measure it at 15 minutes intervals. For a link to be considered congested, the threshold must have been exceeded at least once on 5 or more different days of a 14 consecutive day period.
Presumably this 14 day figure is a rolling window.
So, if its measured to be congested at just a couple hours in a 2 week period, its considered congested for that whole month?
And only a few percent of links actually met this ridiculously low criteria for congestion?
I agree with the first couple of posters, the CRTC has all the info it needs to make a ruling yesterday. What's the freakin' holdup here? |
|
yabos
join:2003-02-16
Ingersoll, ON
| reply to HiVolt
said by HiVolt :The numbers... Can I be the first one to say... WTF? Congestion? Soo, 2-5% congestion, so THROTTLE THE WHOLE NETWORK. That makes lots of sense. |
|
dabbadooo
@bell.ca | reply to dabbadooo
Sorry, I guess I shouldn't take a bio-break in the middle of a post. While I was out of the room, Zinc made the same observation as I did.  |
|
dabbadooo
@bell.ca
| reply to HiVolt
The figures in the table that HiVolt posted go from to Mar 07 (pre-throttle days) to May 08.
I don't know much about networks, so I may be missing something, but if they're saying throttling is the solution to the "problem", why does their massive throttling effort show no net impact on these figures?
Also, do they provide figures for the rest of the country or did they cherry-pick Ontario & Quebec as the worst case (suggesting the real problem is not keeping up with growth)?
Does this represent all Ontario and Quebec or just the worst areas? |
|
mlerner
Premium
join:2000-11-25
Nepean, ON
 ·Rogers Hi-Speed
·TekSavvy Solutions..
·Bell Sympatico
| reply to R0CKY
said by R0CKY :said by HiVolt :The numbers... Can I be the first one to say... WTF? Congestion? Just realised... How can they put a "total" to those values... They don't relate... They all have different thresholds and mechanisms to deal with packet-loss or load. They'd need to talk on a case by case, not on a global... Some of this is on a customer to customer issue other portions are on a more aggregated level.... Excellent points, you should make sure these points are relayed in CAIP's reply.
--
"If bullshit was money this guy would be richer that Bill Gates." - quote by olebiker on Mirko Bibic |
|
R0CKY
TSI Rocky
Premium,VIP
join:2005-05-19
Chatham, ON
| reply to HiVolt
said by HiVolt :The numbers... Can I be the first one to say... WTF? Congestion? Just realised... How can they put a "total" to those values... They don't relate... They all have different thresholds and mechanisms to deal with packet-loss or load. They'd need to talk on a case by case, not on a global... Some of this is on a customer to customer issue other portions are on a more aggregated level....
--
TSI Rocky - TekSavvy Solutions Inc. |
|
sibisties
join:2002-10-19
Montreal, QC | reply to Mirko five_perct
I love the last graph that shows "cell loss" ! That's not caused by congestion, it's their DPI boxes that are dropping packets ! |
|
Mirko five_perct
@videotron.ca
| reply to HiVolt
said by HiVolt :said by Mirko five_perct :
Hivolt, per the publicly filed report:
"While these numbers may seem low to the average lay person, they are significant to network traffic engineers".
k?
I read that... I don't buy it. k? ditto... K! |
|
R0CKY
TSI Rocky
Premium,VIP
join:2005-05-19
Chatham, ON | reply to HiVolt
Still begs the question of the ethernet side... Bell's network is being converted to it and most of their investments have gone to it.
--
TSI Rocky - TekSavvy Solutions Inc. |
|
HiVolt
29
Premium
join:2000-12-28
Toronto, ON
clubs:
·TekSavvy Solutions..
·Bell Sympatico
| reply to Mirko five_perct
said by Mirko five_perct :
Hivolt, per the publicly filed report:
"While these numbers may seem low to the average lay person, they are significant to network traffic engineers".
k?
I read that... I don't buy it.
k?
--
,,!,,('-'),,!,, |
|
Mirko five_perct
@videotron.ca
| reply to HiVolt
said by HiVolt :The numbers... Can I be the first one to say... WTF? Congestion? Hivolt, per the publicly filed report:
"While these numbers may seem low to the average lay person, they are significant to network traffic engineers".
k? |
|
