Sensing mounting frustration that movies Linux ISOs aren't downloading fast enough, the IEEE has announced a new group that aims to bring wired Ethernet speeds up to 1Tbps by 2015 and as fast as 10Tbps by 2020. The announcement comes shortly after the publication of the IEEE 802.3 Ethernet Bandwidth Assessment report, which …
So in summary
As prices drop, adoption of faster technologies increases and there appears to be a need for a premium product to replace the current one as the current premium product tends toward commodity prices. Got it.
I can't say that I care much about Ethernet speeds as long as all I can get is a 7mb/s ADSL that in reality tends to give me 2mb/s.
Then you're not their target.
Their target is datacentres, large networks etc. where a single Gigabit connection is no longer sufficient and 10G isn't going to last long once it gets to sensible prices.
Hell, a lot of people have Gigabit connections at home now, via things like 802.11n wireless routers that need more than 100mbps to make best use of them. And even home users have to copy stuff across the local network that takes time.
But this is really aimed at servers in hosting environments, where 100Mb has been a standard for so long it's laughable when you consider ADSL speeds. Even the cheapest VPS will end up on an 100Mb connection nowadays.
And when you run a network of any size, your switches will start off at the most basic with 10/100 ports (now almost exclusively running at 100Mbps) and Gigabit interconnects (whether copper or fibre). This is the sort of £100 off-the-shelf model that goes into every from a small dentist surgery up. When you outgrow that (which is easy to do if you have even a handful of users, and bog-standard servers now come with dual-redundant Gigabit ports that can usually be reconfigured to load-balance so you can have "2 Gigabit" to the local network) you can multiply that up - 10G is really too expensive and 1G is insufficient for such places at the moment but that's likely to change. Even your average school will have a Gigabit backbone with the servers on it and will run multiple servers, so before you even start you've hit the theoretical capacity of the backbone and would really use 10G if you were doing heavy file-moving or things like, gosh, video editing which is part of the standard curriculum for some subjects.
So 10G, when it comes to sensible prices, will slowly come in (it is already - hell, I could buy it today, it's just a little pricey for what it is) and guess what that does? That makes 100G necessary. The 40/400 G ones are really just temporary measures because of the technology at the time. I imagine only the top-end of users ever touch them and only because there's nothing else standardised. By the time it comes down to the "small LAN" level requiring that sort of speed, they'll skip the 4-whatever speeds and go straight to the next power-of-ten.
So, it's quite sensible to suggest looking into getting things standardised ready for that. Hell, what speeds do you think your local cabinet / exchange talk at? Chances are your data spends most of its transit on a gigabit-capable network at the very least. Just because UK telecoms have yet to catch up doesn't affect what those ISP's are using, what even small businesses are using, what datacentres are providing by default, or what your local network runs at. Hell, my local network was running at 100Mbps back in the days of 56K modems. Your Internet connection speed is only one item out of many uses of your computer (and is an entirely different matter - nobody puts Ethernet to the home yet).
But even there - by the end of the year, lots of people will be able to get 120Mbps home broadband on fibre with Virgin, for instance. Not to mention other fibre providers. And as soon as you do that, that old 100Mbps networking kit starts to look tired and becomes a bottleneck (i.e. your sister copying large files to her laptop over Ethernet will knock down your Internet speed because there isn't enough "spare" on the local network).
Routers for fibre products already come with Gigabit ports by default. Give it another five-ten years and that might be Gigabit. And ten you have to think about what happens to all those companies who have Gigabit coming out of their routers, shared by dozens or hundreds of users who are all using the local network heavily. And that means 10G is only a stopgap until these standards are approved.
While GigE doesn't matter much for "Internet" type stuff, it can certainly be handy for home networking. It's nice to be able to throw things around the home network at 100MB/s and to do it with cheap commodity parts and industry standard interfaces. If they can figure out how to bump that up to 100GB/s then all the better.
The really interesting bits of Thunderbolt minus the hype.
Gigabit is not all that fast
HP blades have come standard with dual 10Gb ports for several years now. Google, when it serves a neighborhood full of 1Gbps connections needs must have some means to aggregate it. Fortunately these higher speeds don't require a rip and replace. The upgrade to the electronics uses the exact same fiber for 100Gbps that it uses for 1Gbps. Transcontinental fibers work the same way. 1.6Tbps isn't a standard, but it is commercially available now and proprietary solutions work if you own both ends of the cable.
I'm glad that newer ethernet standards continue to come out.
But, are those speed figures accurate? I mean, I've never seen a single 10gbps link in operation, and I work with computers quite a bit. Well... wait... I guess I was going to post about how I'd expect 10gpbs at data centers, peering points, etc., but expect loads of 100mbps and gigabit links used at homes and small businesses; but, I suppose in many cases the home ethernet cables have been replaced by wifi.