US Broadband Speeds
I just read a rather lengthy discussion on reddit about current US broadband speeds and how they’re lagging way behind most all of the asian and EU markets. They quote the average broadband (download) speed in the US as 1.6 Mbps. Japan – 61 Mbps. South Korea – 45 Mbps. Sweden – 18 Mbps.
At first glance, this looks like a huge disparity. After thinking about this for a minute, though, I started wondering what the big deal is. Sure – it would be awesome to be able to say that I had a 61 Mbps internet connection to my house, but honestly, how much of that 61 Mbps are you able to use? At home, we currently have 6 Mbps cable internet service from comcast, and at work, we’re on a full-duplex fractional DS3 (10 Mbps) from TWTelecom. No matter whether I’m at home or at work, I’m rarely able to saturate the connection. 99% of the time, I find the bandwidth bottleneck to be on the other end of the pipe (or somewhere in between).
Another thing to add to this discussion: a huge majority of the content on the internet is hosted in the US. This means that when some dude in Tokyo wants to surf Youtube over his pimpin’ 61 Mbps internet connection, the traffic has to traverse the Pacific over cramped fiber trunks – he’d be lucky to get 2 Mbps over those connections.
So – my question remains…why the big push for faster broadband speeds? Until hosting companies and the internet backbone carriers upgrade their network to push more data, it really doesn’t matter.
Am I being naive or is this somewhat correct?
I just really enjoy the fact that you said “pimpin’” in this entry.
Hehe – I’ve been trying to work that one into my daily vocab ;-)
Actually, my search revealed that this entry was my first time using the word.
I think the bigger complaint is the horrible upload speed. That said, while there are many that are limited by the remote end, many intentionally by connection. There have been several times where a download would max at about 10Mbps, but if I opened another download from the same host I would get another 10Mbps. Because of the nature of the connection of a single remote system (or a cluster that has a limited number of nodes), there’s only so much it can serve at once, and it’s often better to limit and be somewhat consistent. That said, the last time I went to download an iso file I was pulling it at 25Mbps (over plain HTTP). If only more places would adopt things like Bittorrent for legal downloads though.
Wow – 25Mbps huh? I assume that was @ Bethel, not at home, correct?
Yea. I wouldn’t need it often, but having that at home would be nice sometimes.
I’m planning on putting a large hard drive in my parents’ computer, and backing all of our important data over the WAN to their location for an offsite backup, regularly. It will essentially be a daily differential. I’ll probably end up iterating through the main directory, zip each folder (if it’s been updated), encrypt it, and copy over the new data.
Another option is backing this all up to my webhost. Since it’ll be hundreds of gigs of data, a really fast connection would be helpful, though it will run in the background and as you said, the bottleneck is likely somewhere in the middle.
rsync (over ssh for encryption) does wonders for remote backup. I do a backup of my web hosting to a local volume each night, and even on the 5 minute traffic graphs it’s only a short 20Kbps blip. Combined with cp -l on the destination volume allowing transparent browsing of each day, it’s an awesome solution.
As for the initial copy, it was a solid 6Mbps (my home connection max) for an hour and a half. Given that, I’m leaning towards the limitation often being on the ends rather than in the middle.
Yes – ditto on the rsync+ssh. I actually use rsnapshot, which is essentially rsync+ssh wrapped in a nice easy-to-use package. It does “smart” incremental backups. I have mine set up to run backups at 6-hour intervals throughout the day. It keeps 4 of those in the rotation. Then it does a “daily” backup – keeping 31 of those, and then 12 “monthly” backups.
I hadn’t realized that rsync copied deltas of the actual files themselves. It would seem there would need to be quite a bit of file data traversal to do the comparison. I’m not running my *nix server anymore, so that makes it slightly more kludgy as well, but it may be worth it. I’ll have to investigate. The issue for me is I want the backup itself to be encrypted, not just the transfer of it. (And if it’s encrypted I don’t need to bother with ssh anyway)
Well you don’t want your user/pass going across the interwebs in cleartext, now, do you?
Regarding file deltas…I believe if you’re transferring from linux->linux, it knows how to transfer just the deltas, but I got the feeling that when transferring from windows->*nix, it has to re-copy the updated files. That may be incorrect, though.