Tuesday, 18 August 2009

802.11n: pushing wired networks to the edge?

Draft 11 of the 802.11n Wireless LAN specification was published on the 23rd of June 2009 and looks set to be ratified in September. What does this mean for schools implementing wired and wireless networks? Does the promise of faster wireless signal the demise of wired networks and cost savings for school construction? This is a more technical post …

I’ve been reflecting on the wired vs. wireless debates we regularly encounter in BSF and Academies and recalled the battle for market supremacy between Token Ring and Ethernet in the early days of LAN technology. Back then, in the blue corner was IBM (and a company called Proteon) and in the red corner was everyone else. Ethernet was a descendant of a radio technology called ALOHAnet where users would ‘shout-out’ and compete for access to the network whilst Token Ring used a ‘pass-the-parcel’ technique more suited to IBM’s sense of order.

Token Ring was a superior technology; its ‘token-passing’ approach yielded 80% effective throughput under load compared with Ethernet’s measly 40%, and it used star-wired hubs with self-healing trunks. But Token Ring’s cleverness and IBM’s near-monopoly came at a price and that, ultimately, was its downfall. Ethernet chipsets were cheaper to produce and vendors then addressed Ethernet’s other shortcomings using smarter hubs and switches as transmission speeds over UTP cabling increased from 10BASE-T to 100BASE-T to 1000BASE-T and now 10GBASE-T. The rest, as they say is history.

So what does this jaunt down memory lane have to do with the wired vs. wireless debate? Ironically, 20 years on Ethernet might appear to be under threat from another contention-based radio technology; more specifically the ‘soon-to-be-ratified’ 802.11n which WLAN vendors would have you believe – citing transmission speeds of 600Mbit/s - will do away with wires for good. If you’re involved in a school capital rebuild or refurbishment you could also be on the receiving end of pressure from builders to install the minimum of UTP cabling to slash their M&E budget and hub room sizes. “Everyone has a home wireless network, so why can’t a school?”. No-one would argue that wireless LANs are incredibly convenient for home and mobile use but they simply can’t match wired network performance for ICT-rich learning so be prepared to stand firm!

The two fundamental constraints of WLAN technology which still exist in 802.11n are the contention for use of the Access Point radio and the propagation behaviour of RF transmissions. Both factors mean that effective throughput drops with increasing users and distance. 802.11n draft 2 products are already deployed, certified by the Wi-Fi alliance, which suggest that 100-120Mbit/s is the likely real-world throughput for a single 802.11n Access Point. Admittedly more access points and clever proprietary features can improve this capacity and resolve cell design constraints.

With plenty of ifs, buts, and some careful network design a single Access Point can give a ‘class’ of 15 laptops a learning experience broadly similar (at 8Mbit/s) to a decent home broadband service. Is it educationally beneficial though to impose this constraint within a 21st Century school? Probably not. Should schools be aspiring to serving up media-rich content up to 100 times quicker; the difference between a human runner and a jumbo jet taking off? Well, probably, yes. Especially when they’re investing a large portion of the ICT budget in user devices which can exploit Gigabit Ethernet interfaces.

For now wired networks still have the edge in other areas too including traffic control and end-point management. WLAN security is a moot point where WPA2 AES encryption is in use - but still a concern in poorly-controlled environments - and there’s an increasingly vocal lobby pressing for removal of wireless from schools altogether on health grounds.

Fortunately wired and wireless networks aren’t mutually exclusive so the best new school designs have hybrid networks that do both; horses for courses; wireless for convenience and wired for performance. Clarity about the function of learning spaces is the key to getting this balance right; avoiding over-reliance on wireless and over-engineering of wiring. A broad principle is to provide robust ‘blanket’ wireless coverage throughout the building environs and also wire every structured learning and admin space according to expected device occupancy. Additional ceiling cabling allows wireless capacity to be added later as more densely populated spaces mature. Less structured learning zones and open plazas also need wiring precisely because of their intended flexibility and often the best way to provide this is using floorboxes (cue sharp intake of builder’s breath and words like ‘slab’ and ‘screed’).

Will wireless LANs ever kill off wired Ethernet in the way Ethernet did with Token Ring? Well certainly not within the lifetime of the current round of capital investment in schools, especially with Ethernet heading for 10Gbit/s speeds, and not if we have to unplug wireless networks as a result of a Health & Safety scare. I’ll post more on 802.11n deployment, 10G Ethernet and WAN bottlenecks in a later post. My advice for now? Stay wired.

Rich Torr