Burning Question: Does Internet Speed Vary by Season? | WIRED
It used to be that we would notice this, except that it had everything to
do with temperature *and* dampness. In the '90's, it was still quite
common for a lot of older outside plant to be really only "voice grade"
and it wasn't unusual for copper to run all the way back to the CO,
through a variety of taps and splice points. Even though Ma Bell would
typically do a careful job handling their copper, the sheer number of
potential points of failure meant that it wasn't unusual for water to
infiltrate and penetrate. If I recall correctly, the worst was usually
a long, hard cold rain (hey we're in Wisconsin) after which people who
had been getting solidly high speed modem connects would see a somewhat
slower speed.
... JG
Trying to pinpoint the failure point on one of those circuits is a PITA as well. Getting a telco tech out to test on a circuit that only goes down when it rains is an exercise that Sisyphus would probably decline.
Network Engineer, JNCIS-M
214-981-1954 (office)
214-642-4075 (cell)
jbrashear@hq.speakeasy.net
http://www.speakeasy.net
No, I did not read the article . . . But, . . .
Yes, DSL speed varies by season . . . or rather, temperature.
But, this is really only the case for _aerial_copper_plant. Buried plant is nearly the same temperature year round.
Copper pair resistance changes with temperature. And, therefore, the link speed of DSL will change depending upon the time of the year (temperature) and geographic location.
If there is a difference of but a few degrees of temperature year round, then no there will be no difference. But, if you live in the desert southwest or even the mid-west where the temperatures can be 70-120 degrees different between seasons or even 40-70 degrees different between night and day . . . you are going to have pronounced differences in link speed.
Worst cast, your link speed might vary 10-20%. The longer the cable length from the central office, the more the variance will be. But, this is something that must be measured on a case by case basis. And, since much of the aerial plant has been replaced with buried plant, this really isn't much of a problem anymore.
BBC
Joe Greco wrote:
Please don't forget moisture content. DSL speeds may drop during
wet winters because cable pits fill with water. 
Those with real statistics, please stand up. I know ISPs who run
large DSL infrastructures have these stats. I've even seen them
at conferences.
Adrian
I may be having my wires a little crossed (I'm not an electrical
engineer) but I was always under the impression that manipulation of the
physical characteristics like that from heat/dampness didn't reduce the
"speed" but the "quality" (like line noise/errors/etc) of the line.
Whether old telco lines or newer data lines it's all about electrical
signal and bit error rates. More errors = more retransmissions = slower
perceived throughput.
Just my thinking.
Scott
Joe Greco wrote:
Me! During the rainy season of recent past years, the cable vault in front of my home would flood, thereby degrading or completely hosing DSL service. Haven't had heavy rains for a couple years so no trouble.
I had to replace my DSL modem about 6 months ago because the previous Westell Wirespeed modem had died very slowly. My speed went from 1.5M to less than 200k and was flaky. The new modem gives me a clean 3M/768k connection. Not bad for DSL ($35/month). But that wasn't weather related. Verizon.
matthew black
california state university, long beach
Quoting Joe Greco (jgreco@ns.sol.net):
> Burning Question: Does Internet Speed Vary by Season? | WIRED
If I recall correctly, the worst was usually a long, hard cold rain
(hey we're in Wisconsin) after which people who had been getting
solidly high speed modem connects would see a somewhat slower speed.
Matches my story exactly.
I once had an ADSL connection which, on dry periods, synced at the
maximum of 8Mbps and when it had rained for a day this would drop to
about 6Mbps. It always worked though, the SNR just went up. I regularly
rebooted the modem to make it retrain.
An engineer from the telephone company came to check the wiring but
couldn't find anything that would constitute replacing the line or
something less drastic...
Regards,
-Sndr. (NL)
Well, since it's been documented that internet speed / usage varies with
the weather (it gets faster when it's sunny, slower when it rains) I'm sure some
seasonal correlation could be found.
cheers,
--dr
I may be missing a little bit here by jumping a bit in the thread so
sorry.
What is the difference between weather and seasonal?
I define weather like, well its cloudy and raining here and get in the
car and drive 20 minutes and it is clear and sunny. I would call this
mostly localized, like ground zero. Nice here bad 15 miles/minutes
away.
Seasonal, well I think of Seattle, almost always rain/clouds..., more
than a 15 minute/mile radius. Seasonal reminds me more of say, for
months straight the ground/air is well, frozen. Like the northeast,
where I used to live and will never go back.
Just my .02, I'll shut up now.
Could you point to the documentation?
I having trouble with language that sounds like one concept and I suspect is in fact another. Take as one example the basic digital signaling hierarchy. The specifications call for a certain rate plus or minus some number of parts per million. If they are within tolerance, the amount that they would speed up or slow down is measured in a pretty small number of bits per second. So I don't think the speed of the links is materially changing. If on the other hand we are discussing the volume of traffic using that available capacity, it is absolutely clear that there are diurnal, weekly, and seasonal variations as well as growth in time.
Are we talking about bit rate, which one might expect to be modified by environmental characteristics and is in fact very tightly controlled to prevent that, or traffic volume?
Not true with modem type technologies, where the available transmission
rate is a function of how many available frequency space slices are
deemed to be "good" at any one time.
This isn't really like SDH (from what I've read of SDH, anyway.)
Adrian
Having worked in Operations at various ISPs in rain-riddled Houston for 1.5
decades, I can say that when it rains, water gets into the copper lines in
the ground and caused increased copper-based local loop failures.
That experience leaves me open to believe that where the internet backbone
is copper based, when it rains, failures may ensue due to old or improperly
installed outside plant and could cause failures which would slow down the
internet.
I would also conjecture that more people would be on line during bad
weather, so that internet usage would increase and perhaps over-wrought
links (copper or otherwise) could have some congestion.
Finally, in those places where the internet is experienced through wireless
links, some may see weather related slow downs.
Well, since it's been documented that internet speed / usage varies
with
the weather (it gets faster when it's sunny, slower when it rains)
I'm sure some
seasonal correlation could be found.
Could you point to the documentation?
I having trouble with language that sounds like one concept and I
suspect is in fact another. Take as one example the basic digital
signaling hierarchy. The specifications call for a certain rate plus
or minus some number of parts per million. If they are within
tolerance, the amount that they would speed up or slow down is
measured in a pretty small number of bits per second. So I don't think
the speed of the links is materially changing. If on the other hand we
are discussing the volume of traffic using that available capacity, it
is absolutely clear that there are diurnal, weekly, and seasonal
variations as well as growth in time.
Are we talking about bit rate, which one might expect to be modified
by environmental characteristics and is in fact very tightly
controlled to prevent that, or traffic volume?
Well, a friend at one particular large internet exchange says he can predict
semi-accurately the ambient temperature/ weather in the local city from the MRTG stats.
The stats he showed me backed him up - or at least clearly showed
strong correlation between weather and traffic levels.
The formal proof is left as an exercise for the reader. 
This has nothing to do with corrosion and all about usage and congestion.
Cold weather leads to more people snuggling up with their laptops.
In sunny warm weather everyone gets away from the kb and goes outside
to have a real life.
cheers,
--dr