When I was at Georgia Tech, I got 650 Mbps. You realize the bottleneck isn't your connection at that point -- it's everyone else's. Which means a lot of sites still download just as slow. Although the big sites have optimized data centers, so it was pretty cool downloading an entire OS in a few seconds (although I think my hard drive write speed limited that a bit too).
My father works in the fiber optics industry and has told me that if fiber was brought directly to each home, every person would have more bandwidth than they knew what to do with. One thin, tiny fiber can carry an INSANE amount of information. The problem is the processing circuitry that converts these light signals into digital signals. These NICs have a much lower throughput than the fiber itself, but if the fiber infrastructure was already in place everywhere, upgrades would be much cheaper and quicker. (In other words, Google Fiber has easy upgrade potential to 10, 100, ... Gbps).
> every person would have more bandwidth than they knew what to do with.
I think that is exactly the point. That's why they are doing this; Google doesn't know what exactly will happen when everyone in the USA/world has fiber connections, but they do know that incredible innovation will come. What kind of applications will be built? What kind of applications can be built?
Couple this with the increasing computing power inside each home over the next many years, and Google will have control over an unbelievably fast and large network of computers.
I'd be certain most applications are in the weak-to-strong AI arena.
indeed. The scales of computing technologies have increased by many order of magnitude in their history, and it has never been "enough." Bill Gates infamously demonstrated the danger of predicting how much RAM/Clock speed/Bandwidth/etc. is "enough."
The thing I miss most about my internet connection at GT was the incredibly low latency to most everything. Now, this was circa 2000 or so; but, I remember page loads seeming more like loading files locally the latency was so low.
Google today, for all it's speed an simplicity, is significantly slower to get to search results than Yahoo would load them for me in 2000 on a GT internet connection.
Oh yeah! Think: a harddrive writes at what an average of 75MB/s [1], and 1Gbit internet is 125MB/s [2]. (Note, Google currently advertises 'up to 1Gbit up and down').
This means that dropbox no longer "syncs". It's just another harddrive in terms of how it works. Copies to and from just as quickly as your harddrive.
Heck, it seems like if you installed an application to your dropbox, and ran your computer off of an SSD, it seems like you could enjoy BETTER performance than if you had that program installed on a 7200rpm drive and ran it locally...
Can't wait to see what we end doing with 1GBit ubiquity!
First, even if you "installed an application to your dropbox", it would still be running from your local machine - that's just how Dropbox works.
Second, even if it were running the application from Dropbox's server(s) (it seems like that's what you mean?), it would still have to read from their servers - you can't magically eliminate a bottleneck just by moving it to a different physical computer; data still has to be read from disk.
Though a big datacenter seems like the perfect place to spread the bits around so that you can read from multiple physical disks/computers simultaneously to improve performance. This is unlike your laptop where you are probably not carrying multi-disk arrays around with you.
I'd be interested to compare the "seek time" though. Platter drive seek times can be pretty nasty, but I'd be interested to see if Dropbox or some competitor could serve up better with network latency involved.
I'd use DropBox much MORE if I had more than 15 Mbit upstream. If I had 1 Gbit+ speeds, I'd use DropBox for EVERYTHING (5+ TB) and happily pay for it. As it stands now the upload would take far too long to be much use.
yes. They would have to dramatically decrease their bandwidth charges, and they'd have to lower their charges in general (just 'cause the barrier to entry for their competition would be lower) but if anything, having a fat pipe at home would make storing things remotely a lot more attractive.
Think about it; with that kind of a pipe, you could boot your desktop off of a drive in my datacenter, two cities away.
Think about services like OnLive (who even offers Windows in the cloud). We have these today, but if we had fiber everywhere, we likely wouldn't have the dvd drive in the Xbox 360. We certainly wouldn't need BluRay.
Yes they would. Its the same reason ftp servers, backup servers and NAS would exist no matter how fast the connection speed is.
Its about availability, backups and more holistic access control. It would be interesting to see how these services innovate with such high speed connection though.
Yeah, Google's going to have to do a good job telling people their computers and NICs aren't going to cut it. It's amazing when you're limited by your machine instead of the internet!
Have you ever tried to actually get gigabit speeds from the crap hardware installed in most PCs? Most companies use low end realtek or marvell controllers which may, if your lucky, have a good switch and have jumbo frames enabled get you to ~650-700 Mbps on a single large continuous stream. Performance only goes down from there are multiple connections are dealt with. If you want to take a look for yourself download iperf and see what you can actually put through standard consumer NICs and switches. The only NIC controllers I've seen actually get to true gigabit speeds are from Intel, and you still need a decent switch.
> every person would have more bandwidth than they knew what to do with
The same thing was said about floppy disks and RAM in the past. The reality is that the demand expands to fill whatever you have available and still ask for more. Once we have blazing fast speeds like that someone will come up with a crazy idea that will require still more bandwidth and lower latency.
My father works in the fiber optics industry and has told me that if fiber was brought directly to each home, every person would have more bandwidth than they knew what to do with. One thin, tiny fiber can carry an INSANE amount of information. The problem is the processing circuitry that converts these light signals into digital signals. These NICs have a much lower throughput than the fiber itself, but if the fiber infrastructure was already in place everywhere, upgrades would be much cheaper and quicker. (In other words, Google Fiber has easy upgrade potential to 10, 100, ... Gbps).