大象传媒

I couldn't figure out (even looking at the source data) what they were classing as "traffic".
If it's just data through the pipe, then I suspect the Torrenters would account for a large proportion of the night traffic; you'd have to assume web surfers would form a tiny proportion (and a fair few of that would undoubtedly be porn). Still I found it suprising that at 4am, we are still at 40% of peak.

Hi Digital, Traffic in this instance would mean hits to websites not Bittorrent traffic.

I think europeans are just as polite about not surfing over dinner, although Rory your should note that most europeans do not sit down to eat until 8pm (slightly different in the UK). Interestingly, at my employer we see a significant drop in UK traffic on friday evenings between 5-7 (pub Friday we call it).

I think the assumption that all of america is in new york is a rather dangerous one - 5 hours worth of timezones can have a big impact on when people go to bed, get up, eat...

Hello, I am here.

It would be much more useful to see the data broken down by port number. How much is port 80 (http) vs port 25 (smtp) vs something else?

Judging by the content of my spam folder there are a lot of American machines left on 24/7 which are spam spewing zombies. How much do they skew your numbers?

This comment was removed because the moderators found it broke the house rules. Explain.

I'm guessing that Bittorrent traffic etc is included in these statistics, as the amount of traffic is very high at 4 in the morning, which is probably too late for even the most hardcore web surfers. Torrent downloads use a large amont of bandwidth and if there are people leaving their computers seeding overnight then there'll be people in different timezones downloading during the day. This leads to the fact that whilst a cluster of European servers are hosting websites, the hits for those websites could be coming from further afield such as China.

The other factor could be online gamers (which uses a lot of bandwidth) playing at stupid o'clock to trade/battle/whatever with different regions: Europe/America/Far East, or when they can achieve faster connection rates to reduce lag.

Since the way they have measured web usage isn't clear in the blog it is hard to form an accurate and balanced opinion, but if it is simply raw data usage then the above points could all be factors.

I guess I'm pretty standard then. I usually log on to check what the rest of the world is doing around 0800GMT and then again around 1200. Don't usually hit 大象传媒 again until after 2100-2200 then. I am guilty of leaving the PC on all night though, despite the fact it's not doing any surfing.

Note to self: Make sure you have a power supply handy. Already been through one this year already. :/

Sorry Rory but this is slightly missing the point.

For a site like the 大象传媒 indeed your original hypothesis would be entirely correct. Indeed news and general sites do tend to peak at around 7-10am as people digest their fix of news in the morning. There is another smaller peak at around 12-1 (workers lunchtimes) but the rest of the day tails off significantly.

For sites like Facebook, twitter and other more dynamic sites there is an initial peak around 7am (pre-work most likely) and then a big drop until lunchtime, the difference is that usage tends to keep building throughout the day, especially in the chool holidays as those without jobs break up the tedium of daytime TV and then those of us who do work get home. Indeed facebook usage tends to continue on and can peak around 10pm at times.

Games sites see an even more later skewed peak in times and can often still be plugging away well after midnight.

Also as you pointed out, Europe is much less a hegemony than the US is in terms of usage times, here in the UK we pretty much mirror the US in usage stats accross the board.

This comment was removed because the moderators found it broke the house rules. Explain.

Well, I didn't read your article until 12:25, so it was sat there waiting for three and a half hours for me to stop work and sit down to read stuff over lunch.

Basic journalism: You need to specify the scale being used on the left hand side of the graph

Basic commonsense: The graph goes up to 100 and at 12am traffic is at 90% in North America.

1. At 09:34am on 19 Aug 2009, digital_elysium wrote:

I couldn't figure out (even looking at the source data) what they were classing as "traffic".
If it's just data through the pipe, then I suspect the Torrenters would account for a large proportion of the night traffic; you'd have to assume web surfers would form a tiny proportion (and a fair few of that would undoubtedly be porn). Still I found it suprising that at 4am, we are still at 40% of peak.

-----

We're at 40% of peak at 4am because people know that's the only time you get a decent connection speed here in the UK.

I'd like to see a graph with the time axis on a relative scale to see when the countries' internet traffic peaks instead of relative to local time of day. This way it might explain why I get bandwidth pinching when I do.

While it's interesting, this data simply doesn't really provide anything new beyond what we could already have taken from common sense.