Hello, this is my note about this issue, take it as you will. This also fills a request Andre sent me 3 weeks ago via another thread concerning the same issue. I am no network engineer but I do know my way into it a bit. Please feel free to correct me at will.
As far as I can tell, the issue seems to be from either a bad load balancing on their servers and/or a system that throttle single connections to prevent over-usage of their network by just a few individuals.
The server(s) IP in question here is "220.127.116.11" (Hostname = "ggc.teksavvy.com".
As seen in the video along this post, one single connection while trying to stream at 1080p gives me 1MBps of download speed which in most cases is enough to buffer the video enough to live-stream it but if it was to slowdown at any time then the video would have to buffer, thus creating the issue we are having right now.
Then I tried to download the video from 8 connections from the same server. Each connection would range from 300KBps to 800KBps and cap my download speed. Indicating that this specific server is not lacking bandwidth and/or overloaded at all as it CAN send the data without any issues.
Now as far as I'm concerned, I'm watching videos all evenings for severals hours. It's my main way of passing time while waiting for my WoW raid to start. I am now resolved into just downloading them all and then watching them so I don't have to deal with the buffering.
Can someone from TSI check this out because we have had this problem for a very very long time and it needs to be fixed now!
(P.S... both my external and internal IP are shown in this video. I don't care so don't talk about it)
You're not suggesting that Rogers is throttling TPIA customers are you? That's absurd! We can't just call it "node congestion" and hope that it fixes itself? So what if my connection is crippled from 20:00 - 12:00 nightly? That doesn't mean that Rogers would actually be throttling.
In all seriousness, I actually switched from Rogers, I switched in October. I used to be a long time Teksavvy DSL customer until dry loop costs became too expensive for me. That aside, these "hiccups" which I'm paying $70 / month for, started for me the day I switched from Rogers to Teksavvy. Now, it could be coincidental but I wouldn't call it node congestion unless my entire street signed up for cable internet between October and now.
I'm not sure if I feel like paying $70 for 10/1 service anymore.
update: some changes were made earlier this morning to shift traffic around a bit.. this should help. we are also expecting an other 10gig link late this week and moving traffic to it next week. -- Marc - CEO/TekSavvy
Just thought I'd mention that this past week (starting on about Thursday), I've noticed slow downs and high pings during peak hours in London. Hadn't experienced that yet but I guess it finally caught up to London. Not as bad as other areas, 10~15Mbps with 50-60ms ping on speedtests.
The links are updated as of this morning. I just checked and there is no congestion anywhere right now. Something else is going on if you are still having YouTube or speed issues. -- Marc - CEO/TekSavvy
In Ottawa on Richmond POI and haven't had any problems or lowered speeds. On the 3Mbs/256kbs plan and get that consistently. Speedtest shows 8-9 Mbs due to SpeedBoost but the connection will settle down to 3Mbs after a while. No problems with Netflix, Youtube or other sites stuttering. (note: I limit upstream for QoS reasons so it is lower than the plan's listed 256kpbs)
Any congestion up in Newmarket? I did a consistent ping for about 2000 packets during peak time and was getting about 5% packet loss. Also ping times ranged from 16ms to 700ms on Google, and Skype was terrible. Attached are my current line stats which I assume are in acceptable values since the SNR seems okay.
Guys, I had a look at the network yesterday and one backhaul link in actually pretty congested. It swings back and forth between various backhaul links due to the POI links being spread out across multiple backhaul links.. I.e. some days one back haul link is perfectly fine and other days its not. But never more than one at a time but sometimes it's extra bad... Like last night. So that means that any POIs that have one or more links on that particular backhaul link on that night will have issues. I also believe that this may be part of what's going on with the YouTube situation since so,e packets are arriving out of sequence and stuff like that as they get spread across multiple links. We have an upgrade due next week that I believe should help and we also have another 10gig link due next week or the week after. With that we will be able to sort this out. Really sorry about this folks.
Any update on implementation of the expected upgrades. Peak hours are still pretty rough