Despite WLAN speed growth, do we really need mGig?

Understanding mGig and its WLAN speed implications

When it comes to wired Ethernet, these discussions are usually a lot simpler. We generally run a workstation on a dedicated 1 Gigabit Ethernet switch port, with ample headroom for typical work operations. The network interface card is 1 Gb, the switch port is 1 Gb, and most typical workstations -- not servers -- just don't max out that equation very often. But WLAN and 802.11ac are different. The .11ac standard allows for an almost 7 Gbps data rate, with the current draft .11ax standard promising rates closer to 10 Gbps.

Given that we're talking about Wi-Fi, these numbers are impressive. But they are also misleading. When it comes to wireless, data rates are usually roughly double any actual potential speed because of the half-duplex nature of wireless -- and other factors not worth visiting in this article. 

Then, there is the fact the top-end rates prescribed in the wireless standards often never get realized. The 802.11n spec, for example, promised a WLAN speed data rate of up to 600 Mbps. Yet, it topped out at 450 Mbps -- due to hardware -- resulting in a 25% drop in performance even under the absolute best conditions.

Usually, we all connect in the range of good to very good on well-designed wireless networks, but the top end is pretty evasive in the real world, where we have little control over a litany of factors working against achieving the fastest Wi-Fi speeds.

Defining need and the best route

You don't absolutely need mGig just because you are moving to whatever comes next for APs.

So, back to mGig and the question of need. Alas, there just isn't a clear answer. Making things more complicated, not every environment is the same when making the analysis required to decide. We know it's completely reasonable to assume 2.5 Gbps or 5 Gbps uplinks to APs are necessary. But many of us who have been on .11ac for years now, even in Wave 2, can see that most of our busiest APs don't come close to even momentarily pegging their Gig uplinks. With today's mix of clients, I have 26,000 devices hitting 4,000 APs, and my overall throughput into the controller involved peaks of 6 Gbps to 7 Gbps. This doesn't really cry out for mGig -- yet.

I try to be mindful my lens on the wireless world is hardly all-inclusive. I have an enormous BYOD paradigm in play, and I have no choice but to support some fairly crappy devices that drag the averages down when measuring what's happening in the WLAN. I can envision a more tightly controlled corporate environment where top-end client devices running heavy applications do -- or could -- occasionally push the limits on the uplinks of well-designed .11ac Wave 2 Aps. And, in that case, mGig could have appeal.

We are at an odd place in time for enterprise Wi-Fi. Vendors' marketing departments came out strong when .11ac was new, weaving fairly ridiculous tales of what the new standard would do, thus creating a WLAN speed climate where mGig needed to happen so the wireless and wired stories were in balance. Meanwhile, folks like me have boiled off the marketing hype to reveal 802.11ac really doesn't need mGig in most cases. That said, mGig is here.

If I gaze a few  years ahead, mGig does make more sense to me than it does now. If I have to buy switches expected to last five to seven years and possibly support both current Wave 2 and future .11ax APs -- and the next couple of generations of client devices -- then mGig is a reasonable future-proofing measure to me for my own situation. If your WLAN just isn't that busy and your switches are fairly new, I might advise waiting at least one upgrade cycle to get into mGig.

Only you can decide when the time is right for your organization to start using mGig switches. Whether you take the plunge or put it off should be driven by a thorough analysis of AP uplink utilization and an educated guesstimate of future demands. But you don't absolutely don't need mGig just because you are moving to whatever comes next for APs.