i've got an E1 circuit that goes:
Toronto -> Amsterdam (fiber)
Amsterdam -> Islamabad (satelite)
Islamabad -> Karachi (fiber)
for the last couple days, my router has been seeing an abnormally high number
of "carrier transitions".
a parse of the log shows 400+ UPDOWN syslog messages from the cisco for
today (4:pm Eastern), and some 400 for yesterday.
prior to that maybe 0-10 a day.
i've openned a ticket with the circuit provider, but by asking the list, i
was hoping to be prepared for a "must be solar flares" excuse.
We've been having problems with solar flares, clock speed and electromagnetic
radiation from satellite debris. I told our people to power surge the
drivers though, and now everything's ok.
what does "power surge the drivers" mean?
It involves placing a reverse quantum bias on the line drivers. This
causes an enormous electromagnetic flux which purges out any stray bogons,
which are often the cause of mysterious route flapping.
if you are gonna use XML, at least use it properly.
From a terrestrial perspective, the recent solar flare activity would have a much more noticeable effect on circuits that have long, inductive (copper) loops that collect those pesky electron charges. In North America, I imagine that this is mostly local loop issues, since the majority (everything?) that is of any appreciable distance is fiber. It is perhaps the case that the Canadians and Alaskans may see more of an issue even on their short local loops, since the more polar segments of the Earth bear the brunt of the storms (if I recall my astronomy courses correctly.)
(Extra credit: I know that solar flares cause more havoc the further North one goes. The British Isles/Scandanavia about the same latitude as parts of Quebec, which has had past problems with the power grid and solar flare activity. Do any IP providers in those areas (NA and EU) build solar events into their contingency plans? Hint: drag this conversation to the datacenter mailing list.)
The relative (not complete) immunity that the North American networks see may not be the case in fiber-poor areas such as Pakistan, where despite your carrier's claims of fiber it is very possible that you're on copper for quite a distance, more than would be expected in NA.
Were you asking about satellite or terrestrial issues? It's very difficult to determine from your comments if you think that this is a satellite or "other" issue (I'll assume you know it's not the fiber.)
As a datapoint for satellite: Over the many, many hundreds of Ku-band satellite earth stations that my current employer runs, there have been no measurable (or at least statistically significant) changes in our error counters, over several continents, latitudes, and satellite/transponder sets.
Ku is quite a high-frequency spectrum range; lower ranges will see more errors and noise due to the nature of the physics, so I can't speak for those people who push IP across those other frequencies.
But none of this matters, since we all know that the Internet was designed to be hit by nuuuucleahr bombs and keep on workin', right? What's a few billion trillion electrons to the Internet?
>i've got an E1 circuit that goes:
>Toronto -> Amsterdam (fiber)
>Amsterdam -> Islamabad (satelite)
>Islamabad -> Karachi (fiber)
>for the last couple days, my router has been seeing an abnormally high number
>of "carrier transitions".
Were you asking about satellite or terrestrial issues? It's very
difficult to determine from your comments if you think that this is a
satellite or "other" issue (I'll assume you know it's not the fiber.)
my apologies, i should havc pointed at the satelite link.
however, i have heard of solar flares being respnsible for various
communications and electrical interference.
i used to manage a circuit to the canadian arctic, which was (as i was told)
sensitive to "sun spots".
there was also an incident where quebec hydro had some serious failure that
was blamed on solar activity.
But none of this matters, since we all know that the Internet was
designed to be hit by nuuuucleahr bombs and keep on workin', right?
What's a few billion trillion electrons to the Internet?
actually, the link is still "workin'", however, due to routing flaps every
5 minutes or so, the customer has a different opinion on "workin'".
considering this started happening in the last couple days, without any
configuration or hardware changes, i was looking for explanations.
BTW: the circuit provider has found an abnormal number of CRC errors, and
is going to run some end-to-end tests when we can coordinate a time.