Once again, I went through my logs pertaining to the stability of my ResNet connection for the month of October. Overall, the connection was down more, but there was one whole day (out of 31) where no issues appeared. It also experienced two major outages.
For October, there were 109 outages of varying lengths, including a three hour outage on the 7th and a roughly seven hour outage on the 13th/14th (more on this one below).
There was also a roughly hour long outage (it wasn’t out for the whole hour, but it was too sow to be usable) early in the morning of the second, as well as large latency spike in the afternoon of the eighth and some during the day on the 19th.
The internet did work properly on the 11th, though this was likely due to reduced load do to many students not being in residence on account of Thanksgiving.
I was not actually at residence on the 13th, being home for the weekend and in transit from home, but I was able to monitor some of the situation from home via my Shaw connection.
The issues actually started on the 12, with noted high levels of packet loss and general instability appearing in the late afternoon and early evening. The ISP was notified, but likely due to the long Thanksgiving weekend, wasn’t able to do anything about it.
The 13th saw some of these issues clear up, but in the late afternoon, things started degenerating again. This lead to the internet going completely down around 6PM and staying out, with a few small exceptions for another 7 hours!
Additionally, routing to SFU, which seems to be hard-coded on both routers at the end of the route (which makes absolutely no sense to me) was not working until that evening, and it still continues (as of right now on November 3rd) to still not work properly.
I did inquire as to why the internet went down and spent so long down, and I was told that it was on account of an attack on the router. Why this was not fixed on the 12th when the issue first started cropping up has not been answered yet, and neither has why it took seven hours to fix a preventable problem.
Personally, I run software on my router (Snort, among others) that is really good at detecting and blocking potential attacks. I don’t know why the router wasn’t properly protected.
The internet connection was up a rather low 97% of the time, meaning that it was unusable for almost an entire day spread over the month!
Additionally, transfer rate was lower this month than last, with an average downstream speed of around 14Mb/s (versus ~17Mb/s for September) and around .75Mb/s upstream (versus .85Mb/s for September). Variation throughout the day increased further, with speeds dropping to around 5Mb/s during the slowest, peak hours.
Average ping on the internal network this month also crept up by roughly 13%.
Based on my observations from this month, I think my conclusion from last month that the network just isn’t designed to handle the load it is under is correct. The day of relative stability when lots of people weren’t here seems to indicate that when the load is low, stability is increased.
The large outage also highlighted the need for proper equipment, which represents a big factor in stability.