Problems persist this morning on the internet front. I woke up to find web pages taking upwards of 40 seconds to load. And I was hoping to work today…
What’s interesting is that, today, raw speed doesn’t appear to be a problem. It’s the packet latency that’s causing problems. Check out this test from this morning:
You can see that download speeds are improved, but the Ping time is abysmal. I would normally expect a ping time under 100, or about a tenth of a second; that 3282 ms ping time is over 3 seconds. A poor ping time often points to a routing issue. Which, even when there’s good throughput, can negatively effect delivery of packets to a client.
It’s like, even if you have a really fast car, if the road to your destination is poorly defined or too circuitous, no amount of speed will get you there quickly.
And, indeed, something is going on in regards to the route that packets are travelling in and out of Whitehorse.
I did a traceroute to the server that hosts Northwestel’s website in San Jose, California. (What? you didn’t know Northwestel doesn’t trust its own facilities for web hosting? Hm. Sorry to be the one to burst your bubble this morning.).
I found that there is trouble right out of the gate at a server in, of all places, Yellowknife. Check this:
Check out line number 3. I get an abysmal ping time of 3884.472 milliseconds from that router. That’s over 3 full seconds.
That server, IP address 22.214.171.124, is listed as being located in Yellowknife. Which is weird because, holy cow, I thought there was this amazing fibre optic pipe that went through BC that we were all on.
Anyway, as you can see in the traceroute above, there’s a block of 5 routers that all have absolutely terrible ping times: 4114.462 ms, 4049.268 ms, 4745.705 ms, and 3761.709 ms. Those routers are located in Yellowknife, Yellowknife, Toronto, and Toronto, respectively. At least that’s where Northwestel and Bell have registered them as being located.
My guess: Northwestel’s illustrious fibre is still down. It’s been down for days. And the company is routing us all-the-way-for-the-nearest over through Yellowknife through a group of routers that, for some reason, are very unresponsive.
And, while the bandwidth may be okay, the routers along the way are either half asleep or overloaded with the additional traffic. And that’s causing the slow-down.
So what does this mean in the long term? Is the fibre so damaged that it requires a long-term fix? Maybe there are just so many renegade backhoe drivers down there this summer that it’s not a workable solution this season. Are we to suffer a summer of internet traffic that follows the route of John Franklin? Northwestel, then, best be wary of cannibals…
Methinks that maybe, just maybe, the design of the fibre route is so FUBAR that we’re in for a long summer of slow internet in the Yukon.