802.11ac phase2 and copper links

  • 1
  • Question
  • Updated 5 years ago
  • Answered
given 802.11ac will start to deliver more than 1Gb (and is often fed at only 1Gb) phase2 will possibly offer 2.6Gb (and more depending on MU-MIMO) - what is Aerohives take on the mechanism to deliver that kind of data to the APs themselves - 10Gb copper (shortening cable runs, moving cabs into satellite locations?) or fibre to the AP (FTTA? ;-) ) - even 10Gb to the edge switch seems inadequate for the deluge of data that can be driven at the edge..40Gb seems to be just enough for a switch driving 24 APs 8-)
Photo of alanbuxey

alanbuxey

  • 2 Posts
  • 0 Reply Likes
  • unsure

Posted 5 years ago

  • 1
Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes
Perhaps we need to let some contention apply rather than unduly worrying about the worst case and see how it gets used in the real world? Then consider what is the appropriate contention ratio based on the environment.

For me, I certainly think we need to see:

1Gb/s switches with 10Gb/s uplinks falling in price so that hey're more cost effective at the edge than they are today. You can get 20Gb/s via link aggregation.

10Gb/s switches with 40Gb/s uplinks falling heavily in price for aggregation and core purposes. (Hopefully Broadcom's Trident II will start to to bring the cost down to a sane level as it starts to get built in to newer generation switches.)

I don't think there is a clear answer today...

I suspect we might well see SFP+ ports on Wave 2 APs giving us deployment flexibility.
Photo of Jenni

Jenni

  • 93 Posts
  • 7 Reply Likes
Do you ever sleep? Thanks for being with us from start to finish. Now that's dedication!!! Oh, and I'll get Matthew to chime in to help you out......;-)
Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes
I've been happily coding all day and it has been a very welcome distraction! :)
Photo of Matthew Gast

Matthew Gast

  • 284 Posts
  • 63 Reply Likes
We discussed this a bit in the question 6 thread: http://community.aerohive.com/aerohiv...
Photo of Matthew Gast

Matthew Gast

  • 284 Posts
  • 63 Reply Likes
To be more complete: fiber is a non-starter because it doesn't supply power. One of the big advantages of PoE is that it doesn't require a licensed electrician to validate the install because it's low voltage wiring. (The city of Redmond, Washington, home of Microsoft, does have a permit fee for installation of low-voltage wiring, but they're an exception.) Fiber for connectivity and electrical wiring for power is two cables that can't generally be done at the same time; two cat 6 cables can be put in at the same time by the same contractor.

For the time being, there are two viable options. One is to require dual 1G connections to an AP. The 3.5 Gbps number for wave 2 requires that you have a 4-stream device with 160 MHz channels. Any time you back off the max MCS rate, 4-stream operation, or the full 160 MHz channel, the speed goes down. I doubt you'd see sustained bursts in excess of 2 Gbps. (Also keep in mind that wireless is half-duplex, so a port group connection serves 2 Gbps in each direction.) The other is that many chip manufacturers have 2 Gbps extensions for copper, so the choice might be "use two cables for vendor-neutral deployments, or one cable and we'll use the special FastGigabitEthernet mode on our switch."

And yes, like Nick, I want to see the cost of the silicon components of edge and aggregation switches drop, since the future of Wi-Fi depends on it.

If we tried to do power over fiber, the switch would need a really powerful laser so that the AP could convert the laser light into electricity to power up. But then we'd need this sticker on every AP (or at least its fiber connection):
Photo of Nick Lowe

Nick Lowe, Official Rep

  • 2491 Posts
  • 451 Reply Likes
PoE for power, SFP+ for data?
Photo of Craig Mathias

Craig Mathias

  • 63 Posts
  • 0 Reply Likes
Well, there's always wireless electricity. Just kidding. That won't work either.