Cisco Cisco 5520 Wireless Controller White Paper
1 |
P a g e
Executive Summary
Dramatic improvements in wireless networking combined with an increasingly important role for
mobile devices is driving a trend towards Wi-Fi as the primary network access mechanism in many
networks. An increasing number of enterprise networks and especially mission critical networks like
healthcare cannot afford to have application sessions drop at any cost. These trends are driving
CIO’s and network managers to demand increasing levels of service resiliency. Wireless network
vendors are addressing these needs by implementing new capabilities, more intelligent wireless
networks, that are able to prioritize mission critical traffic and dynamically recover from component
failures. High availability (HA) services are not new. Vendors have long supported key HA capabilities,
including dynamic RF management to mitigate AP failures or interference as well as dynamic failover
of AP’s to backup controllers, should a primary controller fail. However, while these first-generation
failover services have decreased the duration of service outages from hours to minutes, they have
failed to meet the test of user transparency. Applications often time out and users are usually forced
to reauthenticate once an AP has failed over to its backup controller. To meet service standards,
enterprises and other vertical industries need more sophisticated failover services that are stateful in
nature, presenting themselves as a nearly imperceptible delay in network access for users and
complete restoration of application sessions.
Cisco approached Syracuse University’s Center for Convergence of Emerging Networking
Technologies (CCENT), an applied technology research lab with 15 years of experience testing Wi-Fi
products, to perform a systematic beta test, including before/after benchmarking of several
applications, of their newest Wireless LAN Controller (WLC) Software (code version 7.5), which
includes a new controller failover feature called Client Stateful Switchover (Client SSO). We tested this
new service offering on our wireless testbed that included Cisco 5508 Wireless LAN controllers and the
Cisco AIR-2602i access point. Client SSO represents an evolution in HA services for Cisco’s Unified
Wireless Network Access architecture. For many years, Cisco supported stateless failover of AP’s to a
backup controller. In August 2012, they introduced stateful AP failover capabilities to speed up the
failover process, and while most applications would recover in 3 to 4 seconds, client state was not
maintained and re-association was required. The release of v7.5 builds on this by promising stateful
failover of not just AP’s but client state as well. This new Client SSO service copies AP and client state
information from the primary controller to the secondary controller’s memory on a constant basis in
the background. The promise is imperceptible application downtime on the client side in event of a
primary controller failure. What previously took 3-4 seconds to occur now takes place in milliseconds.
Key Findings:
mobile devices is driving a trend towards Wi-Fi as the primary network access mechanism in many
networks. An increasing number of enterprise networks and especially mission critical networks like
healthcare cannot afford to have application sessions drop at any cost. These trends are driving
CIO’s and network managers to demand increasing levels of service resiliency. Wireless network
vendors are addressing these needs by implementing new capabilities, more intelligent wireless
networks, that are able to prioritize mission critical traffic and dynamically recover from component
failures. High availability (HA) services are not new. Vendors have long supported key HA capabilities,
including dynamic RF management to mitigate AP failures or interference as well as dynamic failover
of AP’s to backup controllers, should a primary controller fail. However, while these first-generation
failover services have decreased the duration of service outages from hours to minutes, they have
failed to meet the test of user transparency. Applications often time out and users are usually forced
to reauthenticate once an AP has failed over to its backup controller. To meet service standards,
enterprises and other vertical industries need more sophisticated failover services that are stateful in
nature, presenting themselves as a nearly imperceptible delay in network access for users and
complete restoration of application sessions.
Cisco approached Syracuse University’s Center for Convergence of Emerging Networking
Technologies (CCENT), an applied technology research lab with 15 years of experience testing Wi-Fi
products, to perform a systematic beta test, including before/after benchmarking of several
applications, of their newest Wireless LAN Controller (WLC) Software (code version 7.5), which
includes a new controller failover feature called Client Stateful Switchover (Client SSO). We tested this
new service offering on our wireless testbed that included Cisco 5508 Wireless LAN controllers and the
Cisco AIR-2602i access point. Client SSO represents an evolution in HA services for Cisco’s Unified
Wireless Network Access architecture. For many years, Cisco supported stateless failover of AP’s to a
backup controller. In August 2012, they introduced stateful AP failover capabilities to speed up the
failover process, and while most applications would recover in 3 to 4 seconds, client state was not
maintained and re-association was required. The release of v7.5 builds on this by promising stateful
failover of not just AP’s but client state as well. This new Client SSO service copies AP and client state
information from the primary controller to the secondary controller’s memory on a constant basis in
the background. The promise is imperceptible application downtime on the client side in event of a
primary controller failure. What previously took 3-4 seconds to occur now takes place in milliseconds.
Key Findings:
1. Time sensitive applications like Microsoft Lync, Citrix VDI and Adobe Connect experienced
imperceptible application downtime in the event of a controller failover event
2. The failover time in v7.5 is 98% less than that in v7.4 which only supported AP Stateful
Switchover.
3. The failover time in v7.5 is 99% less than that in v7.2 which supported only state-less failover.