The future of AC (Wave2)

[L&LD]

Could use. Emphasis on could.

I have to disagree with your assessment overall.

The AC Class specs clearly state what performance is to be expected of a fully realized product. With AC Class Wave2 devices Wifi will transition to effectively an wireless switch topology, vs. the archaic hub architecture we have been stuck with so far for our wireless networks.

2 or more 1GBe ports with LAG is not an option. Not when with even 10GBe ports a mere 3 fully MU-MIMO AC Class Wave2 clients with multiple antennae will fully saturate the network.


I am starting to think a little like Tim here at this point. Unless and until we get products with fully realized specs, the latest router that makes sense to buy for me or my customers is an RT-AC68U.

The RT-AC87U could be a contender too if the price gets in line with what it actually offers.


Having a fully functional Wave2 router with 4 antennae or more, 160MHz channel widths and support (and clients) for MU-MIMO but without 10GBe ports on board is like paying for a race horse, but one that is permanently attached to a buggy.

http://www.buggy.com/Images/websluggy.jpg


Nice to look at, but the old horse and buggy would be just as useful, if not an even better match.

 

[sfx2000]

It will be interesting to see if device makers give this a shot. Wireless USB never made it and Wi-Di / wireless display still isn't in wide deployment, at least not in a form that is widely interoperable.

True - Wireless USB was fragmented by issues within IEEE, along with the different business alliance groups - good technology in there, and a lot of money was burned, unfortunately - I have friends that worked on the UWB side of things - it was good tech...

WiFi Direct - this pretty much died on the vine, at least in the initial context as it was not widely adopted, and even then, it was beat up fairly badly by poor implementations. Maintaining two wifi links on the same client adapter is tricky, Apple is finding out the hard way right now with their implementation on OSX 10.10/IOS8.xx - and that is in a walled garden, and it's still hard to do right...

So 11AD might be the best shot at it - seems to be more focused than the other approaches, and more vendors are on-board with it.

 

[azazel1024]

Unless you get really crazy with 8 stream designs, you are looking at a realistic max of 2.7Gbps raw wireless rate with a 4:4 160MHz design. With MU:MIMO, you lose effectively 1 stream, which actually bumps it down to roughly 2Gbps if you have a number of MU:MIMO 160MHz clients connected.

Figure in error correction overhead and the BEST you can possibly hope to achieve is around 1.7-1.8Gbps...which means that a pair of GbE ports in LAG actually more than handles the possible throughput that the router could see.

The only instance where LAG is unlikely to ever make sense is where you are encountering >4:4 stream routers or 3:3 or 4:4 clients that are also utilizing 160MHz (2:2 with 160MHz might also >1Gbps, but it is going to have to be a VERY good connection to be able to exceed that).

For a client device, I think those scenarios are going to be a lot more limited.

One other assumption here is that the router and any switches or other devices that are connected to it are not going to be able to split the TCP packets between interfaces, so each client is going to be stuck with no more than 1Gbps, just that multiple clients can be aggregate across the links, just like how LAG generally works now (as opposed to being able to do multichannel, which is much more a client sided thing as far as I know, and not a network device thing).

Anyway, 10GbE would of course be the ideal, but since I think we are a good 3-5 years (isn't everything?) away from that being even typical prosumer/high end enthusiast gear, I don't see router manufacturers (except enterprise) entertaining that. LAG on the other hand is solely a firmware feature and pretty danged easy to implement (granted you'll need devices that support LAG behind the router too, but again, not very expensive and easy to implement), both in the home and ESPECIALLY in any kind of business.

LAG could realistically be deployed now on all new routers and most home/business users could take advantage of that at either no cost, or possibly no more than a $60-200 additional cost (if not less on a per installation cost for multiple installations).

A final note is, this is all assuming very good connections for all clients. Since it probably isn't likely that you have 2, 3 or 4 clients all in the same room, all that have very high throughput capability (11ac 160Mhz or not), you might really be connecting a bunch of clients that are each only squeezing 20-40MiB/sec, even on an 8:8 router design as most are a room or two away, or line of sight, but a good distance, etc. So you are back to LAG being more than enough in a lot of cases.

 

[RMerlin]

Ok. If it connects via USB that would make one 4x4 client. But not a very portable one and more of a dedicated bridge vs. a compact client useful for mobile use like notebooks.

Agreed. And already, 3x3 clients are quite rare in laptops, so I wouldn't expect any internal 4x4 client to appear anytime soon. So you're probably correct that 4x4 (and future 8x8) applications will mostly be used for bridges, with very few niche clients such as the EA-AC87 out there. And as I just remembered, the EA-AC87 isn't just a client, but is in fact a bridge (assuming the product hasn't been radically changed since the original announcement by Asus - such delays might indicate radical changes).

 

[L&LD]

Unless you get really crazy with 8 stream designs, you are looking at a realistic max of 2.7Gbps raw wireless rate with a 4:4 160MHz design.

Of course I want to get crazy with 8 stream designs.

When such a router (true Wave2 AC) is used as an AP, the low port requirements that you state go out the window.

Especially when we take into consideration the wired clients it should be supporting too.


No problem that each client is (and will be for a long time) limited to 1GBe connections. But it is a problem when the router internally has more power than it can actually give to the connected clients.

This is becoming more and more an issue with ISP speeds hitting 1Gbps up/down - and the routers ports are saturated just by routing, let alone simply switching between internal LAN clients too.


Thinking that LAG is sufficient even today is a little short sighted.

With True Wave2 AC**, it makes even less sense.



** True Wave2 AC is a router with enough processing power (ram, cpu and nand) to fully exploit 160MHz channel widths. MU-MIMO enabled. 4 or more antennae. The logic to enable wireless as a Switch instead of a Hub as it currently works right now.

 

[azazel1024]

Honestly by the time that vision is fully realized, 802.11ax is likely to be rolling out.

It is going to take 10GbE seeing wide deployment and much lower prices before you are likely to see any consumer router with a 10GbE port, even high end $500 ones with 8 antennas and all the bells and whistles.

Though that said, at the current pace, I'd be suprised if we saw wide spread MU:MIMO deployment before 2015 is over and done with. It doesn't seem like it is an easy feature to sell on the router side of things and nearly impossible as a selling point on the client side. Too much explaining needed.

A consumer can kind of get "Oh, that phone has 802.11ac. That's newer and faster than my current "old" phones 802.11n...right? And the other new phone I was looking is 802.11n also...hmmm. I do want my facebook and netflix to go faster". Qualcomm and Broadcom don't seem to be at the point where their 11ac client chipsets all ship with MU:MIMO (or even most of their higher end ones).

I get the impression we are 3-9 months away from real consumer client products shipping with any MU:MIMO support and none of those, as far as I know, support 160MHz. It took years for 802.11n to be "fully realized"/tapped out once it started seeing draft products. I suspect it is going to take close to the same amount of time before 11ac is as well...probably 4-7 years and in that time, maybe in 3-5 years, we'll probably start seeing draft 802.11ax products begin to trickle out.

Probably only in the later part of 802.11ac's maturity will 10GbE be a possibility, let alone likely. Maybe, possibly, we might see LAG occur sooner, even though in some ways that is harder to explaining to typical consumer, it IS easy to implement and nearly free, other than firmware/testing/feature support (no real hardware costs)...and since you'd likely only see it in higher end consumer routers, that is somewhat more the demographic that just might "get it" and not need much explaining of how to implement or why you'd want to get a router with LAG.

I have zero prognostication ability, but I just have a gut feeling that LAG will happen with at least higher end 11ac routers and probably well before 10GbE is something seen in any kind of consumer router.

 

[L&LD]

I can agree that LAG may surface as a feature before we see 10GBe ports on consumer routers.

But as 11ax is estimated to debut in 2019 (at the earliest), I think we're more likely to see 11ac fully realized before then. Feature-wise, at least. Like you I can't know if those features will actually work in the real world, at least initially.

But there is no way the manufacturers can continue to give us new products without any real reasons to upgrade for the next 3 years. Unless they work their way towards True Wave2 AC (I should patent that) in some meaningful way. And to me, the most meaningful would be with 10GBe ports along with the other perks of AC class wifi.