This is an antiquated and stupid convention that really only applies to transit carriers. Netflix pays their wholesale ISPs for transit. Retail ISPs pay their wholesale ISPs for transit. Peering allows them both to save that money. The ratio doesn't mean anything because every single bit originates from a paying customer of the Retail ISP. The only Retail ISPs that don't want to peer with Netflix are the ones who also sell video content.
No, it really isn't. This is about Retail ISPs who also sell video (cable companies) wanting to limit access to Netflix for their customers without doing something obviously like rate limiting/filterering at the gateway router. The 'people in the middle' like Level3 and Cogent are bypassed completely when Netflix peers with an ISP, which saves both Netflix and the ISP money. Hint: I work for an ISP, we don't sell video, we peer with Netflix everywhere we are able as well as install their caching appliances in our offices to reduce our transit and backhaul costs.
If I can go 1000 miles, Dallas to Chicago being a real world example on my network, then I can go 1000 kilometers and then some can't I?
This equipment is being deployed by carriers and ISPs, and generally carriers and ISPs have been complicit in the surveillance with the "Five Eyes" anyway, so this isn't a big purchasing concern when buying from Cisco or Juniper.
In fairness, this is layer 1 stuff that they can't really backdoor. I guess they could create a big red "shutdown" button in Beijing though.
The routers are what you have to worry about forwarding select interesting traffic back to the mothership.
I still wouldn't use Huawei transport though, it's honestly not that cost competitive with home grown vendors like Infinera and Ciena.
That's no acheivement. 1tbps over 1000 miles is easy. That's ten OTU4 channels running over a simple DWDM system. All the coherent hardware out there has great OSNR performance so running through 10 amplifiers over 1000 miles of OSP fiber is an everyday project.
What this article is about, what everyone is working on in the optical space, is a single 1tbps super-channel. Everyone's already concluded we can't hit 1tbps on the same 25ghz spaced channels we use for 100G, so they are working on concatenating 12.5ghz slices into a larger superchannel which may be 50, 62.5, 125 ghz spaced, whatever the needs require. However, that's hard to understand so the blurb is just 1tbps over 1000 miles, which sounds impressive to someone who hasn't already been doing that for years.
Because Coherent optics are wavelength-specific on the receive side, you could set up a 40ch or 80ch system with nothing but 1:N splitters. The problem there would be the 1000 km reach discussed in TFA.
We've been turning up 1tbps optical transport for years, this is easy. You can do this with commodity parts. What they've probably done, which isn't in the summary or TFA, is turn up a single 1tbps super channel over a flexible grid ROADM. That's currently in the development stage with a lot of vendors, such as Alcatel, Ciena, Infinera, Cisco and more. That would allow the entire ROADM system to scale up the N-Terabits, where N is going to depend on how many superchannels can be crammed into the C-band. Probably on the order of 50-100 terabits per second fully loaded.
Have there been? 10,000 people die in Japan from a tsunami and everyone is still shitting their pants over a nuclear reactor that didn't kill anyone. People fear what they don't understand, and most of us aren't physicists.
Because it's an nVidia product and it's sole purpose is to break compatibility with AMD hardware.
You're talking averages when the only thing that matters is peak. At 8AM and 5PM +- an hour all taxis and all commuters are on the road at the same time.
It just shows that at a certain level of volume it makes more sense to produce the product yourself. These companies need to move terabits of traffic long distances. If they did this purely by leasing capacity from traditional telecoms they would be paying millions of dollars per month. For that kind of money, you can build one hell of a transport network and then have control as well as cost savings. Unfortunately with Submarine routes the only reliable way to get fiber in the cable is to be a member of the consortium building it. You can't come along years later and buy dark fiber because it's all in use quick.
Correlation is not causation. Living in a high crime area motivates people to defend themselves.
I grok the snark, but in my experience people who take their own initiative on learning and personal development gain 10x more than people who get sent to some boot camp or seminar on their employers dime. If you are learning something useful it all comes back to you in a future paycheck anyway. "I haven't been trained on this" is generally an excuse I hear from people who wouldn't know their ass from a hole in the ground even if they did attend a 3 day ass-recognition boot camp.
Seems like a solution looking for a problem to me.
People said the same thing about the laser for decades.