Stop pissing about and give us consumer 10Gb/s now
at a reasonable price.
Make no mistake: the flood of data flowing across networks today will grow into several Niagaras. We can foresee zettabytes of information crossing the internet – and the lion's share will flow via Ethernet links. Without Ethernet our networked world would cease to exist. Without Ethernet development the internet will choke. …
at a reasonable price.
Its coming. Thunderbolt on consumer-grade intel boards, followed by lightpeak this year, (apparently).
Surely someone will write an IP driver for it.
I wonder if you have dual ports you can daisy chain computers together in a ring and use some sort of token-passing mechanism for network access...
Don't worry, I'm going...
Now we are beginning to see the 10G PHYs with reasonable power requirements, they should become ubiquitious and hence cheap.
One of the biggest bottlenecks (for certain things) isn't the network, it's disk write speed.
I've done analysis work on many networks and we struggle to capture all the data on a 10Ge link, let alone a 40 (assuming it's running at 50% or so).
Distributed architecture for disk access needs to be developing as fast as the network if this growth rate is to be sustained.
I'm pretty sure the right SSD raid system would handle data rates like that just fine, you'll just choke on the price for it, then again, if you're running 10/40Ge you already have a healthy budget. Also it makes no sense to try to flood a single node with 10Gb traffic and expect it to catch it all, surely tens if not hundreds of nodes are generating the traffic. Have the capture node split the traffic to $x number of devices and use software smarts to aggregate what you are looking for.
I need a lint brush now
"Terabit Ethernet will surely go the same route, from technology dream through standardisation and production to everyday reality"
in 20 odd years. 40Gb is bearly a reality now.
It already is a reality - check out geant and Ja.net these days!
That's a bit of a stretch. I was vigorously supporting Ethernet when many people were claiming that Token Ring or ATM was the way to go. But without Ethernet, we'd just be using a different (perhaps less effective) layer 2 protocol and the Internet (TCP/IP) would carry on just the same.
Ethernet's become merely another brand these days. The protocols that deliver 10G (let alone 10T) bear only a passing resemblance to what we used to run over banana cable. Can I have a 'long grey beard' icon, please?
I can't believe it. A whole artiicle like this and comments too, without mentioning IP.
Not that I disagree very much, but you can be sure that wouldn't have happened if Cisco or juniper had been involved.
Of course if it's the speed of the pipes you are talking about then it's almost bound to be Ethernet - although some people would also mention SDH or even OTN.
Pretty useless speeds when they can't even deliver a steady 10mb to the punters already paying outrageous prices.
Plus the usual bandwidth caps and tiered pricing, that should do the trick.
There is something horribly wrong with this article but I can't quite put my finger on it.
It's got more fluff on it than a poodle in a tumble dryer, that's why. Ponderous grandiloquence ("Without Ethernet our networked world would cease to exist.") coupled with meaningless statements ("Businesses and other organisations have chosen Ethernet for networking applications because it has become cheaper, more interconnected, more reliable and more manageable than its alternatives").
None of which is to demean Ethernet. It's a great technology, and transformative in the way that very few technologies are (USB is another good example). And plugfests like the EA are a great idea. But I don't need breathless journalism that reads like it's been written by Barbara Cartland amped up on Mountain Dew to tell me that. Just the facts, please, ma'am.