(sorry about lack of threading etc, I clipped the text out of the web-based archive) Jonathan wrote:
David Robinson wrote:
"For a start, it takes more than two fibre cuts in the
North Island to
disrupt us"
And also remember back early in the year when one slip on the Hutt motorway took out all services in Wellington.
What is so wrong with Telstra's network that single events can affect so many people, when supposable it takes more than two fibre cuts to disrupt them?
This post you've quoted was related to the core network - i.e. the bits that connect Wellington to Auckland etc. This is a fault in the access network i.e. the bit that connects Mary and Bob (or in this case Kandallah) to the core network. NO operator has full redundancy in the access network for residential services.
What amazes me is that Telstra maintain that a truck hitting a pole was a *completely* unforseeable event and that even in retrospect they could not have been expected to have forseen that *one* *day* a truck would hit a 'single-point-of-failure pole' and that they could not have been expected to have planned for it. I ask you, is that really reasonable? I thought that engineers were the types of folk would look at diagrams of such a network and say: "well this pole is a single point of failure and its right by a road. Sooner or later its going to get hit by something, we'd better plan for that". I don't get the feeling that TC *had* planned for it... though maybe they did and if they hadn't planned for it we'd be disconnected till the new year... Insights anyone? Also, is there anywhere which keeps a log of outages experienced by NZ ISP's, how long they last etc? Thanks
What amazes me is that Telstra maintain that a truck hitting a pole was a *completely* unforseeable event and that even in retrospect they could not have been expected to have forseen that *one* *day* a truck would hit a 'single-point-of-failure pole' and that they could not have been expected to have planned for it.
I think that the question one must ask first is "Just how many such poles are there?" Probably thousands scattered around the country. Just the same as there are thousands of point where a backhoe can cut fiber and disrupt *local* networks. That is what the telcos mean when they say "that they could not have been expected to have planned for it". It has been repeated pointed out in this discussion that residential subscribers do not pay for redundancy. At work we have two connections to TC via two different fibers going from two different buildings over different paths and, of course we pay for it. That's in addition to a connection to APE and additional international bandwidth from there via KAREN which is completely separate for the TC circuits. This is backed up by duplicate routers, firewalls etc. and plenty of redundancy within our network. Ultimately these are all 'security' decisions. To make such decisions on a rational basis you weigh up the risks against the cost of failures and the probability of them happening and the cost of fixing them. There is *never* enough money to remove all the risks so you then fix those offer you the best return. Even if you are spending 'enough' for some value of enough you will occasionally get caught out by low probability events (or faulty analysis ). Accidents will happen. Count on it. Russell
Russell Fulton wrote:
What amazes me is that Telstra maintain that a truck hitting a pole was a *completely* unforseeable event and that even in retrospect they could not have been expected to have forseen that *one* *day* a truck would hit a 'single-point-of-failure pole' and that they could not have been expected to have planned for it.
I think that the question one must ask first is "Just how many such poles are there?" Probably thousands scattered around the country. Just the same as there are thousands of point where a backhoe can cut fiber and disrupt *local* networks. That is what the telcos mean when they say "that they could not have been expected to have planned for it".
"local networks" is inherently a relative term :) This one affected several suburbs.
It has been repeated pointed out in this discussion that residential subscribers do not pay for redundancy. At work we have two connections
Redundancy isn't really what I mean; more forward-thinking. From what I've heard out of Telstra people -- and thats just the faults and 'helpdesk' people, the attitude seems to be that a 'truck hitting a pole' is not something that could have been foreseen and that therefore there was no point in having a contingency plan to deal with it. To me, this just seems wrong-headed. My question for Telstra is, was there a plan? How well did it work out? If they didn't have a plan or if it didn't work out well, how are they going to address this in future? And the answers I've been getting back from Telstra people are rather disconcerting.
At 08:39 p.m. 22/12/2006 +1300, Steve Wray wrote:
Russell Fulton wrote:
What amazes me is that Telstra maintain that a truck hitting a pole was a *completely* unforseeable event and that even in retrospect they could not have been expected to have forseen that *one* *day* a truck would hit a 'single-point-of-failure pole' and that they could not have been expected to have planned for it.
its called a restoration plan and they have them and it worked very well.
I think that the question one must ask first is "Just how many such poles are there?" Probably thousands scattered around the country.
millions - you have no idea of how many towns and cities have single points of failure on poles. And thats world wide. In the mid 70's a single lightning strike hit a pylon and caused 5 days of rioting and looting - in New York. The power was off for days. The pylon carried 5 circuits and loosing them destabilised the power network and it fell over like a pack of cards. Rebooting a power system like New Yorks isn't a 3 finger salute. So it comes down to risk and chance. What is the risk that you'll get hit crossing the road ? it never happens ? So do you always use an overbridge or tunnel ? We all learn to live with a degree of risk. I know customers at the end of Telstra's fiber. They're not happy, but neither do they want to pay more for diversity. So they have a strong cup of coffee and accept the dice rolled the wrong way.
On Fri, 2006-12-22 at 17:22 +1300, Steve Wray wrote:
What amazes me is that Telstra maintain that a truck hitting a pole was a *completely* unforseeable event and that even in retrospect they could not have been expected to have forseen that *one* *day* a truck would hit a 'single-point-of-failure pole' and that they could not have been expected to have planned for it.
I ask you, is that really reasonable?
In defence of our friends at TCL.. In a world of infinite budgets, yes. In the world of reality, no. Single Points of failure are like 'whack a mole' in some respects. They're always there, and all you can really do is shift them to a place of least damage and least probability. In the scheme of things, some plonker with an (illegal?) oversized truck is in the realms of extremely improbable.
Also, is there anywhere which keeps a log of outages experienced by NZ ISP's, how long they last etc?
The first step is standardising on the format that we inform people. Aggregating things to form reliable stats after that is easier. I opened my mouth about something like that on NZNOG a while back and then rapidly ran out of time to actually do something about it. Que Sera Sera. But it's on the backboiler, so at some point something like you mention may come to life. Hmm.. Maybe we should get Uma Thurman on the case. jamie
participants (4)
-
Jamie Baddeley -
Richard Naylor -
Russell Fulton -
Steve Wray