Wireless networking has never been an arena for absolutes. There’s always wiggle room, a list of exceptions, and the “under lab conditions, but will be different in your environment” factor. To the uninitiated, it can sound like we’re either trying to make excuses or that we suffer from the inability to commit when we can’t promise discreet quantity (35 users should all get 12 Mbps at 75 feet from this access point, unless any one of these very likely things is in play…). To our our fellow Wi-Fi professionals, this frequent moving tartgetism is just a way of life that we both accept and pride ourselves on being able to bring order from as we ply our craft. The wireless half of WLAN has always been fraught with permutation, but prior to 11ac, the wired uplink was straightforward. Now that we’re well into 11ac’s tenure, we’re finding that even the notion of planning for getting APs connected to switches has gotten potentially confusing- and the WLAN industry isn’t exactly helping itself in this regard.
The Confusion Is Understandable To A Point
Where managers and non-techie money folks are trying to plan for future WLAN expenditures, you can appreciate the assumption that big, big capacity uplinks might be needed for a new wireless standard that promises to around 7 Gbps. Forget about the “data rate versus real throughput” paradigm for a minute- 7 Gbps is data center-grade connectivity in the minds of many, and so it’s no surprise that people map available Ethernet speeds to what it would take to support the promise of 11ac. Remember here that 802.11ac, as with 11n before it, is WAY OVERMARKETED as ambitious glossy goes right to the we-may-never-get-there high end of the standard. Under that lens, and combined with innocent ignorance of the nuances of real-world wireless, you can sympathize with those who think “hmmm, 100 Mbps ain’t gonna cut it. And standard Gig ports are way too slow. We better plan for 10 Gbps per AP.”
Thankfully, this incorrect conclusion is fairly easy to walk ’em back from.
After Ruling Out 10 Gbps Uplinks, It Gets Uglier
So we get past the point where 10 Gbps is being chatted up for AP uplinks, and we get closer to reality. But in this case, reality seems to be in the eye of the beholder, and there are lots of beholders with their own realities. Unfortunately, they also happen to be many of the same folks that customers turn to for technical guidance in these issues. Right now, about all you can safely say is that the WLAN industry agrees that for 11ac, 100 Mbps uplnks are too slow and 10 Gbps uplinks aren’t needed. Beyond this, it’s pretty wild and woolly. Different though leaders have different opinions, and as bizarre as it seems, they all sound viable. Oy vay.
The short version: given all of the variables of the contemporary complex business Wi-Fi setting, many environments won’t be able to achieve aggregate demand of 1 Gbps or higher even on the latest 11ac hardware. Or maybe they will. But they won’t, and you can count on that. Except where you can’t. So all you need is a a 1 Gbps uplink. But you better run two cables. And burn two switchports. But you don’t need to. And because 1 Gbps won’t be enough (or will it?), a new class of switches is being developed to put multiple Gigabits of throughput on a single UTP run.
<OK, breathe deep… In, out… there. Feel better?>
Yes it’s all a bit crazy. And those perpetuating the craziness likely mean well, they just don’t seem to agree on what’s really “needed” when asked by customers how to cable for 11ac going forward. That lack of unified message really does a disservice to customers in a number of ways:
- 11ac is frequently overmarketed. There is a delta between promise (or implied promise) and what reality will be.
- We’ve seemingly entered a period where everyone accepts “oh, that’s just marketing- let an SE or VAR explain what this REALLY amounts to”
- I don’t think that some in the WLAN industry get that cabling isn’t trivial in many buildings, and even a single cable run can exceed the price of a top-end AP in many cases. Pathway concerns are huge where conduit is in use, and some of us have to get our cable designs right to serve many, many years.
- This status quo makes the industry look a bit disjointed, and kinda silly at times. Wireless is complicated, sure. But a common message on how to cable for it shouldn’t be.
What They Said On The Topic In 2014
…what many people don’t know, is that second-wave 802.11ac APs will require two, not one, Gigabit Ethernet ports. That just doubled your need for switch ports and cable runs. Oh boy!
…11ac is a radical change; if you go by emerging WLAN guidance on prepping for and implementing the latest wireless standard, your to-do lists get significantly complicated.
The short version: 11ac will require two switch ports and two cable runs per access point. Simple AP uplinks now become port channels. Port channels need careful configuration, and can be a nightmare to troubleshoot should one of the four RJ-45 connectors involved with each 11ac port channel get cocked or not sit straight in its port.
In the first wave of 802.11ac, a single 1 Gbps link is sufficient. Wave 1 is 1.3 Gbps, but that includes the substantial 802.11 protocol overhead and is a bidirectional number because 802.11 is half-duplex. For any new wiring for 802.11ac, I’d put in two cat 6 cables for maximum flexibility going forward, though.
Cat6 versus 6a isn’t what’s important, it’s getting two cables into the cable plant. The second wave of products will potentially reach 3.5 Gbps, so you’ll want sufficient backhaul capacity to accommodate that. I wouldn’t stress about the exact specification; just make sure you have two cables that can support Gig Ethernet plus power.
Stressing about the new 802.11ac standard seems to be the industry’s new pastime.
Now that Wave-1 of 802.11ac is here with vendors promising 1.3 Gbps in 5 GHz, 1.75 Gbps aggregate per AP, and world peace, suddenly the industry has focused in the potential bottleneck of AP backhaul links. In other words, is a single Gigabit Ethernet uplink enough for each AP?
The answer is just plain “yes,” and applies not only to Wave-1, but also to Wave-2 11ac…
The IEEE 802.11ac Wave 1 standard has already delivered 1 Gigabit wireless speeds to enterprise access networks. Soon, the industry will introduce 802.11ac Wave 2 products that could deliver wireless speeds up to 6.8Gbps
Earlier in October, Aquantia announced its development of AQrate technology—the silicon that enables the delivery of 2.5- and 5-G over Category 5e and Category 6 cabling. In that announcement and in the current announcement of the NBase-T Alliance, the bandwidth requirements of 802.11ac “wave 2” devices were heavily referenced.
There’s certainly plenty out there to confuse, amuse, and ponder on the topic of planning for cabling for 11ac. This is one of those topics that is arguably more of concern for bigger networks and customers with challenging cabling paradigms than it is for others. And it’s also pretty fascinating to see the different takes and spins put on the subject by those in the vendor/VAR space versus those on the customer end (you know… where the dollars are).
One thing is for sure, at least to me- as 2014 draws to a close, we’re no closer to clarity on this discussion than we were earlier in the year, and it will be interesting to see what develops in 2015 as 11ac continues to explode and we see the front end of Wave 2.
I’d love to hear your thoughts on the notion of cabling for 11ac in different environments. Please drop a comment below, and Happy New Year to all.
Lee, looks like you spent too much time locked into contemplating this. Unless you have extreme cases (WLAN PTP links or explicitly requested cabling redundancy) 1 good (Cat6+) cable per AP is your answer – full stop.
Most people will never achieve even 1GBps. And those few who will (and are not ptp links) will benefit from the upcoming 2.5G Ethernet standard that still works off Cat6/6e. It is specifically developed so that you can pay for yet another WLAN overhaul 😉 http://www.networkworld.com/article/2840287/cisco-subnet/cisco-others-pushing-2-5g-5g-ethernet.html
I actually buy this, and have for a while. My point is not “hey, convince me”, it is “hey, the messaging here is really cocked up”…
But your point is strong. Thanks for the comment.
Well, the messaging is, and was, “we want your $/€/£/¥”
So if I’m cabling vendor – you definitely need 2 cables per AP.
If I’m a switch vendor – you need 2.5G Eth.
And if I’m AP vendor – you need either, both or none, depending on what your budget is, how the LAN/cabling part will affect my WLAN budget, whether I’m also selling switches (pretty sure Cisco will be first to market with pre-standard 2.5G Eth APs, that will only be compatible with Cisco switches 🙂 ), and when will I have the next opportunity to sell you an upgrade.
Business as usual…
Aye, well said.
Lee, thanks for “clarifying” the confusion on 802.11ac cabling! I get what apcsb is saying here, BUT you have nailed it on the head that there is no standard for cabling at all. I would like to see some industry professionals/VARs/vendors all come up with your suggestion on how to cable in different environment (from SOHO to high capacity stadiums). Or, am I just dreaming with you? Great blog and hoping that industry leaders will step up to the plate!
Thanks, Glenn- for reading and the comments. I know I’m not the only one of the opinion that this could all be a bit less messy!
1 Gbit/s full duplex will do just fine until I’m convinced otherwise. Running 20-40MHz channels in most of the Campus ensures me of that. 😉
Thanks for commenting, Herr. Most of us that do and minitor busy networks are likely agree to agree for the short term. But if what if you had to budget for cabling for five years out today?
great article and one to immediately share with my employer. Definitely going to be one of the big topics for the coming year I think.
I. Really don’t think that installing 2 x cables to each AP is a good long terk solution for the Industry. For starters there will be few customers who do and therefore any vendor who looks to specify this as a requirement in the same was as for 902.3at power is going to shed customers pretty quick.
But as you say, with WiFi nothing is ever simple. What advice would i give to anyone looking to do a cabling refresh or new build now ? Definitely to go with two cables of whatever standard has been spec’d. Why? Lower risk. The cost of recabling to add in more outlets later would far outweigh the cost to do it “up front”.
Sure, we may never get to over a gig out of ‘ac but then the cabling will be there long after ‘ac is dead and buried.
Jon, thanks so much for reading and the thoughtful opinion. As I mentioned, this may be one of those issues that people with a few APs say “uh, what’s the big deal?” And those of us who measure AP counts by hundreds per building and thousands per campus say “doubling up cabling counts just in case sucks because that can drive pathway costs way, way up in certain building types”. I think this is also why many in the WLAN industry minimize my lament as well- if you are not actually designing, paying for, and running large WLANs for a living, you likely don’t have the shared perspective. (Not throwing stones at you, but talking generally).
The project I have been working from as the basis of my opinion above is about ~500 AP’s in size, across ~25 buildings. (Fairly narrow use case’s, mind) I might actually try and run some numbers to work out what the extra costs were per AP and what it might cost to re-visit the sites and “+1” each AP location. There are many “softer” costs to bear in mind over and above the direct financial impact, (eg, PM overhead) and if you can’t achieve “invisible cable installs” where it all happens overnight from a customer perspective, possible lost productivity.
I get you for sure – doubling up does add cost and quite possibly, needlessly. When the doubling up to future proof was tabled, I definitely wasn’t able to say with confidence “don’t waste the £”. On the large scale it does make for big numbers; the costs of those individual points do add up, but on balance I still think it is a worthwhile investment until there is some clarity on the likely way forward.
Jon, rather than double cabling APs, I would spend my extra cabling money putting in in-fill ac-only (ie 5GHz-only) APs. Given the short range of the highest ac data rates, you’ll get higher performance that way IMHO. The main problem is I don’t know of any vendors who are offering that type of AP yet. As a bonus, with only one radio it would easily fit within 802.3af.
Pingback: WLAN Vendors: The NBASE-T Ball is in Your Court | Frame by Frame