Clustering to bring high availability to Wi-Fi
Trapeze pools WLAN controller resources
Wireless Alert
By
Joanie Wexler
,
Network World
, 06/25/2007
Sign up for this newsletter now!
Joanie Wexler looks at how enterprises can take advantage of wireless LANs and WANs.
- Share/Email
- Tweet This
- Print
The concept of server clustering will soon be applied to wireless LAN controllers.
The idea — introduced last week by WLAN vendor Trapeze Networks — is to bring the same high-availability benefits afforded
by pooled data center server resources to the Wi-Fi environment.
The capability is part of Phase 2 of Trapeze’s Smart Mobile Wi-Fi architecture. Smart Mobile blends centralized management
and services with the option to distribute some traffic switching to access points for improved performance.
The most prominent component of Phase 2 – which won’t be available until year-end – is a new rev to the Trapeze Mobility System
Software (MSS), the smarts in Trapeze MX controllers that will also enable the resource pooling. According to David Cohen,
Trapeze director of product marketing, enterprises can operate any mix of Trapeze MX controllers with the new MSS software,
locate them anywhere in the network, and the devices collectively operate as one big virtual controller with automatic load
balancing between them. The controllers auto-discover one another and new controllers dynamically configure themselves.
APs associated with a downed switch automatically re-map to another available switch “in milliseconds” without administrator
intervention, says Cohen, who declined to be more specific about the failover time.
That setup should enable hitless, in-service software upgrades, too, which is interesting to Larry Griffith, director of technology
operations at Wheaton Franciscan Healthcare, a Trapeze shop in Glendale, Wis.
“Currently, when we do a [routine] upgrade, the APs are down until we finish,” Griffith explains. “With APs automatically
reconnecting to another controller, we should see no downtime.”
The healthcare facility currently runs about 40 MX-400 controllers configured as primary devices and a similar-sized controller
population configured as backups. The backups are always in use but are lightly loaded in case they need to accommodate backup
traffic.
In this configuration, “APs can fail over to a backup controller, but it takes at least a couple of seconds,” Griffith adds.
“Quite a few applications – such as our Citrix applications – can’t tolerate that.”
Smart Mobile 2 “should let us out of that model,” he says, which will also be important to the hospital’s forthcoming 802.11a-based
voice deployment, to roll out next month using Polycom/SpectraLink NetLink 8000 handsets. “Voice, too, would be affected by
such delays,” Griffith says.
Joanie Wexler is an independent networking technology writer/editor in Silicon Valley.
Comment