WIRELESS INFRASTRUCTURE

  • 02/18/2015
    8:00 AM
  • Rating: 
    0 votes
    +
    Vote up!
    -
    Vote down!

802.11ac Wave 2 & The Multi-Gigabit Ethernet Push

Cisco releases Catalyst multi-gigabit Ethernet technology amid dueling vendor standards efforts to support Wave 2 access points without new cabling.

When Cisco launched its multi-gigabit Ethernet technology to support faster WiFi access points last month, it did so without a lot of fanfare. The technology, which Cisco rolled out across several switches, is designed to accommodate 802.11ac Wave 2 gear without the need for new cabling.

The low-key rollout -- at least for Cisco -- comes as two different groups develop standards for 2.5 Gbps and 5 Gbps Ethernet. Cisco, along with Aquantia, Freescale and Xilinx, launched NBASE-T Alliance last October.  Meanwhile, another group that includes Aruba Networks, Avaya, and Broadcom formed the MGBASE-T Alliance.

With 802.11ac Wave 2 products beginning to hit the market and with 802.11ax in the future, enterprises need to prepare for a jump in wireless bandwidth, Cisco and other vendors say. Wave 2 access points will need more than the 1 Gbps Ethernet connections Wave 1 and older generation 802.11n devices used. Using 10 Gbps Ethernet would require an expensive cabling upgrade to Cat 6E, while 2.5 Gbps and 5 Gbps Ethernet ports fill a gap and allow enterprises to use their existing Cat 5E/Cat 6 cabling, they say.

The NBASE-T and MGBASE-T alliances both see 11ac as a major market transition that needs to be addressed. They have the same objectives, but different implementations, Hasan Siraj, senior director of product management for campus switching at Cisco, said in an interview. "The goal is that we come together, and as we go to IEEE, get consensus and arrive at a standard," he said.

Cisco Catalyst multi-gigabit technology

Clearly, Cisco is betting that the eventual consensus will be at least close to the multi-gigabit technology it introduced last month across several of its campus switches.

The new multi-gigabit Catalyst 3840 can be added as a stack member to existing 3840s, which are converged wired and wireless switches. Cisco released a multi-gigabit line card for its Catalyst 4500E converged wired and wireless switch, and released the 3560-CX and 2960-CX as additions to its compact wired switch line. The multi-gigabit technology supports power over Ethernet.

"We are future-proofing customers for a few generations of wireless standards," Siraj said.

Lee Badman, a network engineer and wireless technical lead for a large private university and a Network Computing contributor, said he's pleased with the multi-gigabit concept because it "can address a lot of the challenges that 10 gig is overkill on." He added, "It's a bit of a bummer that, as usual, we see the market leader versus the rest of the pack on a potential new standard."

Jumping the gun?

At the same time, some wireless experts say 802.11ac Wave 2 doesn't require multi-gigabit backhaul links.

"Industry claims of throughput capabilities exceed[ing] 1 Gbps are correct from a theoretical standpoint. However, real-world client mixes on almost every WLAN will mean that backhaul never approaches even close to 1 Gbps of throughput," Andrew von Nagy, a senior WiFi architect, wrote on his blog, Revolution Wi-Fi.

He advised customers only consider greater than 1 Gbps backhaul with Wave 2 in "very specific locations that have a low-density of high-end devices with very demanding bandwidth needs."

In an interview, von Nagy said he wasn't bothered by the dueling 2.5/5 Gbps standards efforts. "The purpose of going through the standards process is to get everyone on the same page," he said.

However, the fact that vendors are pushing multi-gigabit for 11ac Wave 2 illustrates how marketing differs from what's actually needed in customer networks, he said. "If a vendor said, 'You probably don't need this today, but you should be planning ahead,' I'd be fine with that, but it's not going to sell boxes...This is just another driver to increase the speed at which organizations lifecycle their switching hardware," he said.

Long term, there's definitely a need for multi-gigabit at the access layer, von Nagy said, but he cautions customers against making a big hardware investment before a standard is formed; otherwise, they may run the risk of it being incompatible.

Badman said he generally agrees with those who say 1 Gbps is enough to support 11ac Wave 2, but added that 1 Gbps for higher-end 11ac works "against other industry trends of allowing lots of headroom for traffic spikes and surges and future-proofing."

"Some of us have to cable for 10 years or longer, and I like the thought of maximizing the investment in UTP and associated pathway costs," he added.


Comments

Mulit-GigE Necessary?

I agree with the article.  Before jumping on stop-gap standards, I'd wait until the backhaul utilization warrants multi-GbE.  Even then more APs can be added or just jump to 10GbE.

Right now I'd be running Cat 6e for new installs and monitor AP backhaul utilization.  If after five years backhaul is a huge bottleneck, deploy 10GbE.

Re: Mulit-GigE Necessary?

rradina, thanks for your thoughts. I agree in concept but think it will vary greatly depending on the business and how critical its WiFi usage is. As you said, if the backhaul utilization warrants it, some IT departments might need to look into multigigabit as a stepping stone. If you run a hospital thats heavily dependent on wireless, it may be worth it.

Re: Mulit-GigE Necessary?

Two thoughts: I cringe a bit when I read "just add more APs" or anything that makes it sound like running cable is trivial. In some buildings in some cities, "just adding" anything can be complicated and expensive. Low voltage permits, asbestos issues, and pathway concerns make what should be cheap and easy expensive and complicated. Not all facilities were built in the last decade, or even in the last century. In tough locations, cabling is an investment and in these places (or anywhere where replaces large amounts of 5E with 6A isn't palatable) mGig has potential to be a positive force.

-Lee Badman

NBASE-T

As NBASE-T technology can help continue the use of existing Cat 5e/Cat 6 cables, with speeds up to 5 Gbps at up to 100m, it will be a key technology to deploy Wave 2 products without breaking budgets and keeping the existing cable infrastructure.

Re: NBASE-T

While NBase-T and MGBase-T standards formation is beneficial at this point for future planning, early adoption most likely is not. Despite all the vendor marketing surroundin 802.11ac Wave 2 products, almost nil APs will even approach 1 Gbps backhaul throughput in the real world due to client mix and network overhead. The peak data rate for the next few years will be 3.466 Gbps and that is under ideal conditions. In enterprise networks with multiple AP deployments you won't be using 160 MHz channels, and even if you do you will end up sharing capacity between APs operating on the same channels, causing co-channel interference and reducing capacity versuse a single AP on the channel. The most capable APs and clients that will be released in the near term will only support 4SS. This means realistically the highest data rate you're likely to see is 1.7 Gbps using 80 MHz channels and 4SS (this also assumes you are using all the DFS channels so that you can acutally use 80 MHz channels with interference). Add to that that almost every real network has a mix of clients between laptops on the high end, tablets in the middle, and smartphones on the low end of capabilities. When you mix even a few of these devices together the airtime utilization is the limiting factor and there is no way that an AP will push 1 Gbps backhaul. The only way you could possibly come close is if you have only a few high-end clients, say 4SS laptops, which are pushing a really large amount of data, say file transfers of GB size file(s). It ends up being more of a VERY niche solution in specific locations where this could occur. And if this one client would top out around 1.2 or 1.1 Gbps actual throughput anyways, would they even notice if they were limited to just under 1 Gbps? Likely not. 

Go take a look at actual throughput statistics from any AP on any enterprise network. Typically you'll see average throughput on the backhaul of a few 10's of Mbps, with peaks of a few 100's of Mbps. The marketing of peak speeds by Wi-Fi manufacturers is flashy and highlights what's possible in a lab scenario. But real-world use doesn't align with flashy marketing. Yeah, it's more mundane, but networks must be planned and designed around the actual use-case and business requirements of the customer, not marketing. 

There are two things that might change this and make the need for multigigabit more realistic. One is the release of dual 5 GHz radio APs by manufacturers. This could increase the aggregate backhaul from an AP since 802.11ac only operates in the 5 GHz band. Currently, with one 802.11n radio in 2.4 GHz and one 802.11ac radio in 5 GHz it further reduces the likelihood of 1 Gbps throughput. Second, the integration of a third radio support 60 GHz WiGig could warrant backhaul throughput in excess of 1 Gbps. 

Developing the standard now is productive and a good thing. But without any real driver at this point, and without assurance of compatibility with the final standard, I can't see how actually buying these pre-standard products is beneficial. If I were a customer I'd keep my existing switching infrastructure until I have assurance of compatibility with the final standard and have a real driver for upgrade to multigig backhaul. If I have a need for new switches for some other reaseon (new install, need PoE+/UPoE, etc.) then the only way that I would buy the pre-standard multigigabit switches is if they are the same price or lower than comparable alternatives (such as the Cisco 3750, 3560, 2960 lines without multigigabit). 

I also completely agree with Lee that cabling is a long-term investment in many hard to cable facilities. I'm all on-board with getting most life out of existing Cat5e and Cat6 cabling that's in place today. Just don't spend money on switches until there's a real driver. Otherwise you're wasting your money.

Cheers,

Andrew

Re: NBASE-T

Time to reform marketing methods in the WLAN industry... Andrew's comments are dead-on, and completely at odds with the persistant marketing message where insane claims of speeds and throughput at time of sale are the rule. 

 

Says I.

Re: NBASE-T

Thank you Andrew for this interesting analysis. I believe you're mostly right that NBASE-T and multigigabit cable tecnologies won't be around for a while.

But in large facilities with dozens of APs using 801.11ac Wave 2 it could be necessary. In two weeks I'll be in the Mobile World Congress here in Barcelona, and they have to provide high-speed WiFi for 80,000 visitors, including over 3,000 journalists.

The "Fira Barcelona" already have 802.11ac hotspots across the venue. I was able to use the 5GHz network during another conference three months ago, with speeds on the 100's of Mbps. 

Also for large corporate campuses investing in new cable technologies can be productive.

Re: NBASE-T

Good points Pablo -- except I'd argue that large conference halls suffer far more from capacity on individual AP's as opposed to the throughput they provide. Regardless of Wave 1 or Wave 2, the AP will struggle with the number of users connected -- or attempting to connect. The best fix for this is to shrink the AP coverage area and add more AP's.

Re: NBASE-T

Hi Pablo,

I think you just missed the point of my comment. In high-density environments the airtime is the limiting factor due to the mix of low, medium, and high-end client devices, NOT the Ethernet backhaul. Especially since almost all high-density public environments (like MWC) are comprised of a majority of low-end devices like smartphones.

Take a look at the network statistics from MWC last year (note: I had to edit this URL since Network Computing's comment spam system doesn't allow any URLs; please replace "link" with "http" in the following URL.)

link://www.cisco.com/c/en/us/products/collateral/wireless/aironet-3700-series/w...

- 865 access points capable of 802.11n 450 Mbps in both bands (900 Mbps aggregate)
- Actual peak usage of 1.2 Gbps at most during the event
- That's only 1.387 Mbps per AP during peak... far from even the peak throughput of an 802.11n AP which we could estimate around 450-500 Mbps if both radios were maxed out.

Sure, some APs might have been hit harder than others, but not even close to the marketing of peak single-client throughput on each radio under lab conditions.

802.11ac won't change this fact.

Cheers,

Andrew

Re: NBASE-T

Yes, i find this Multi-Gigabit Ethernet Push, we already have 802.11n is available in place today and is sufficient for many customer use cases. But catch here is when to upgrade for 11ac or when i can i expect 802.11ac AP to become available.

Cable costs

In this blog post, Cisco cites an estimated cost of $200 to $1,000 for pulling a new cable.

 

Re: Cable costs

My point of concern is even when we have 802.11ac in place, how many end user devices will be fully compliant with 802.11ac. Or my company will give me new laptop to work.