Hacker News Reader

Comment by jdavis703 | Hacker News Reader

Or each cell site will user a point-to-point microwave relay to avoid digging up earth.

It is still difficult and expensive to do true 1 Gbps line rate (FDD) by microwave PTP, although the $18,000 to $20,000 equipment cost for that can be a great deal less than the $200,000 to dig up the street in a major urban area.

However it is REALLY difficult to accomplish 6 Gbps to 10 Gbps of capacity between two routers via PTP microwave. Whereas it's really easy with two $250 1550nm SFP+ in two strands of dark fiber, if you have an optical path from point A to B.

Or even with one strand of dark fiber using bidirectional 1490/1550 SFP+.

Ubiquity gear around the 1Gbps rate isn't anywhere close to $18-20k.

And for very short links, there are projects like http://koruza.net/ which use standard (cheap) 1Gbps or 10Gbps media converters for free space optical links.

I'd love to know about affordable microwave 10Gbps equipment, I'm not sure we've gotten there yet.

There are cheap P2P MW/NIR solution that can provide high bandwidth but it's still not enough.

But they have downsides, range and more importantly density is limited. there is a limit to how many of them you can deploy before they start interfering with one another even tho they are pretty unidirectional put 4-5 of them between 2 buildings and you can start having problems.

The other part is that they aren't as scalable, the main costs of fiber is laying it but you can lay cables with 50-100 fibers at almost the same cost of laying a single fiber.

Fiber can also be upgraded the same cables that gives us 10, 50, a 100 or even a 1000 gbit per second today gave us 1gig or even a 100mbit just some years ago.

The only thing you need to replace are the transmitters and or repeaters if they are used, again those can be expensive but a lot of metro networking equipment is built to be modular and upgradeable the fibers themselves often to not need to be replaced.

Pretty much every upgrade in fiber optics is built around the common diameter and refractive index fibers no one is going to buy into a new technology if they'll have to rip off all of the fibers and put in new ones.

The US and many parts of the world need better networking, it's about time that cities or at least in the case of the US states from local corporations that would deploy the infrastructure and either contract ISP's to do it or do it themselves and lease the lines to them for future use.

Mimosa Networks is a newish Ubiquity competitor and their cheapest offering for a 750mbps is only $300 for the pair of dishes... Needless to say $20k is a pretty dramatic overstatement.

They have a 1.5gbps offering as well in licensed 10.7ghz bands for ~$3k which will dramatically reduce noise and interference, but that's still well within budget. I think to get 10gbps, your're going to need to operate at a higher licensed spectrum where there aren't many commercial people playing yet.

Telecom grade MW P2P links start at 20K and upwards. They do also however work on licensed spectrums. The new un licenses 24 50 and 60 ghz links can be really cheap, but I don't know how well do they compare to the bigger telecom ones. RF equipment is pretty darn expensive to make, the telecom one have pretty impressive "RF dark magic" eco chambers milled out of aluminum blocks and many other cool features. The new commercial ones seem to be pretty simple and mostly solid state not much more complex looking than your typical wifi router. How do they fair with weather and interference how well do they coexist with each other, and how reliable are they in the long run is another question. Some of those MW drums can operate for decades I'm not sure about the 300$ or even the 3000$ ones.

yeah, no. Not in 5 GHz. Sorry to tell you the marketing hype doesn't jive with reality. Two Mimosa B5C radios and two 32dBi dual polarity high performance dishes will make a nice link, and it is awesome to be able to buy a radio link that will do dual polarity 256QAM modulation in 20, 40, 80 MHz wide channels, but it will not be 750 Mbps full duplex.

In the real world more like 250-300 Mbps full duplex, or slightly faster than that if you tell it to allocate greater bandwidth in one direction in a ratio like 75:25.

You can't compare a Mimosa B5-Lite or B5C to a carrier grade ISP, FCC Part 101 licensed band radio (6, 11, 18, 23 GHz etc) that's fed from dual -48VDC A and B power feeds and has multiple SFP slots in it.

Also you often don't have to dig; mole, duct, blow - there are options.

We have been swapping out duplex sfp's for single fibre working duplex sfp's for a few years on our backhaul links here. It is a great way to increase capacity. We have also rolled out a photonic switching fibre network for customers to lease 100gbit circuits across the country. Photonic switching (basically routing at the optical level) is impressive technology 1 rack can push terabits of data.

However the equipment is in the $1,000,000 range. But that is peanuts for a large carrier.

I've seen photonic switching refer to a lot of different ideas. Sometimes it's just a remotely-controllable add-drop multiplexer.

Does this type of photonic switching differ from CWDM/DWDM?

Are routers putting the signal onto different wavelengths depending on where they want packet to go?

A lot of the equipment I've seen is largely described in broad marketing-friendly terms which makes it hard to understand why the equipment is actually more useful than existing approaches.

No it doesn't differ, its all modulating wavelengths of light through a piece of glass, CWDM using LEDs and DWDM using actual lasers. Routers put packets onto an interface queue at layer 2. Layer 1 what is what handles which wavelength(lambdas)those bits modulate.

CWDM doee actually use with lasers in some contexts. The big difference is that CWDM uses frequencies that are all far apart, with just a few frequencies. DWDM uses frequency channels that are very close. DWDM performs better because it focuses the energy in the band's that have the best properties in terms of absorption, distortion and group delay.

Yes the "C" is for course and the "D" is for dense. All of the CWDM gear I have been exposed to was LEDs, generally actual lasers are much more expensive and you dont need that sensitivity for such loose banding. In what implementations is CWDM using lasers instead of LEDs?

CWDM uses lasers too, the reduced wavelength tolerance simply means that the lasers can be uncooled (reducing the power draw) and that chirp (frequency shift during modulation) does not cause interference with neighbouring channels, making them easier to modulate directly. The 'loose' CWDM banding is actually smaller than the bandwidth of typical LEDs.

Photonic switching is basically a lot of servo motor controlled mirrors in a box, you can accomplish the thing with a human and a lot of SC/UPC to SC/UPC patch panels and cables. It's an OSI layer 1 thing and agnostic of servers, switches. The "box" lets you reconfigure crossconnects in software. It does introduce a degree of loss.

Microtrenching is becoming a thing, which is rapidly dropping the costs of installation. It's just a groove 1.5" wide and about 6" deep. Drop a couple of conduits in it, and back fill it. Vehicles can drive right over it during the process, no need to close lanes for multiple days.

Does that mean every time you replace the road surface you also need to redo the fiber installation?

No, that's how they deployed coax (and later fiber) in Israel when the cable company was formed in the late 80's at a fraction of the cost.

The cables are laid in a trench about 30 CM deep just off the curb the trench itself is maybe only 10 CM wide you have a single machine that digs the trench lays the coax and then covers it with gravel (which is made from the asphalt, dirt and gravel it digs up) after that they just put the asphalt over it and it's done they you can do 100-200M an hour or even more like this pretty easily.

If you need to redo the road you don't have to scrape off the trench part anyhow no car can drive so close to the curb, and even if you do scrape that part off it's just like scraping the road you usually only go maybe 1cm/half and inch deep you don't remove the entire surface and even if you need to redo the road completely you can leave the 10cm wide trench alone.

Also fiber and coax cables can last for decades there is no need to redo them, the same cables can be used again and again with each upgrade to the standard since to get more speed you only need to change the optical or RF equipment there is no need to change the cable.

The same cables that used to deliver 100mbit can be used for 10gbit and even higher speeds, multiplexing is also possible over older cables.

One thing i wondered it why simply they just don't make curb stones (e.g. http://image.made-in-china.com/2f0j00AeqanjdwtkoK/Blue-Stone...) with a few ducts in them already couple inches in diamiter so you could just pull fibers through at any time without having to even dig up the road... Yes you have to replace them, but how often do you that? every 10 years? 20 years? and if you design it well you can either pull the cable out or heck even replace the cable with the curb stone, people can live without internet for 6 hours during the night.

The main obstacle to microtrenching and surface inlaid is not technical , but political. Many cities' streets departments won't allow it. It will take an aggressive education and lobbying effort. It is less costly than cut and cover but in $/km of fiber is ridiculously expensive compared to aerial. Even more so when you see that the limits in strand count for very small diameter cables are 72-144.

Reply on news.ycombinator.com