1 Vidar Stokke Senior Engineer at the Norwegian University of Science and Technology, IT-division, Networking Programme: 1.History of wireless networks.

1

Vidar Stokke Senior Engineer at the Norwegian University of Science and Technology, IT-division, Networking

Programme:1.History of wireless networks at NTNU2.The wireless network with standalone APs3.The wireless network with controller based APs4.Pros and cons of controller based networks

Controller based wireless networksController based wireless networks

2

History of WLAN at NTNU

• 2004/2005: 250 autonomous APs (802.11b og g)• Hotspot coverage

– public areas and meeting rooms

– Approx 20% coverage

• Authentication by web portal (HP) and/or VPN (Cisco)• Challenges:

– Time consuming administration– No mobility– Capasity issues due to many users on few AP– Web portal did not scale

• 1100-1200 associated clients

3

The ”Wireless Campus” project – with lightweight APs• The goal was ”100%” wireless coverage in NTNU’s

buildings.• Started Q3 2006 and terminated Q3 2008• The process:

– Site survey (External company)

– Wiring and AP mounting (3 different electrical contractors)

– PoE-switch installation in wiring cabinets (NTNU IT)

• Resulted in approx 1400 APs, 18 wireless controllers and 100 PoE switches

4

The ”Wireless Campus” project – with lightweight APs• In 2011:

– 1800 APs

– 20 controllers

– 10.000 simultaneous clients at peak hours

– Approximately 300.000 sqm coverage

• Deployment of controller based wireless network was a success

5

A comparison of standalone and lightweight/controller based wireless

• Standalone– Everything on and through the AP

• Traditional lightweight – Everything on and through the controller

• Hybrid lightweight– Almost everything on and through the controller, but with a certain

flexibility

6

Wireless with standalone APs

7

Wireless with lightweigh APs

8

Wireless with lightweight APs and H-REAP

9

AP join process - discovery

• Ways to discover controllers– Layer 2 broadcast on local subnet

– List of previously known controllers

– OTAP (Over The Air Provisioning) – removed from Ciscos SW

– DHCP option 43

– DNS lookup for ”cisco-capwap-controllers.domain”

10

AP join process - CAPWAP

11

Mobility groups

• Several controllers can join one mobility group• Client roaming across different controllers and APs

– Without loosing connection

– Without the need to reauthenticate

12

Mobility groups

13

Auto-RF

• Each controller is part of an RF network– APs connected will forward RRM-info to its controller about other

APs it sees.

– Adjusts channel and power assignment based on leaders computations

• One controller is the leader of the RF domain– Collects necessary RRM-info from other controllers

– Tries to avoid neighbour APs on same channel and interference

– Calculates and computes the best possible channel and power assignment for the different APs

14

Rogue detection and containment• APs not within the RF-domain is marked as rogue

• Possible to do a joint effort to take them down– Up to 4 APs join together

– Send diassociation messages to the connected clients

15

Downsides with lightweight

• Get stuck with one brand (ie. Cisco)– Even though CAPWAP opens for third-party components

• Harder to do hardware upgrades on APs and/or controller– Controller and APs need to support each other

• Expensive with a low number of APs

16

Benefits of lightweight solution

• Easier management/configuration of a large number of APs:– All configuration done at controller and pushed to APs

• Major changes done by the click of a button

– Software upgrade done from controllers and deployed to APs

– PlugNPlay of new APs and replacement APs

• Radio Resource Management (RRM)– Auto-RF

– Client load balancing

17

Benefits of lightweight solution

• Better mobility for clients– Due to mobility groups

• Improved overview of the wireless network– Client information and debugging– AP statistics– AP and client alarms

• Location based services (requires WCS)– NTNU Campusguiden

18

A demo?

19

Questions?