Installation guide

83
Wi-Fi Location-Based Services—Design and Deployment Considerations
OL-11612-01
Deployment Best Practices
number of received responses associated with a larger controller population. Use of a few centrally
located large capacity WLAN controllers (such as the WiSM) would therefore appear to be advantageous
over the use of many distributed smaller capacity WLAN controllers for the same number of tracked
devices.
The impact of these polling activities can be seen quantitatively by examining the protocol analysis
shown in Appendix A—Polling Traffic 2700 <-> 4400 WLAN Controller, page 109. Here you see the
UDP packet exchange between the location appliance and a single 4400 WLAN Controller. To better
illustrate the combined impact of enabling all polling categories, the test controller has been configured
to poll for client, asset tag, rogue, and statistical information every 60 seconds. Seven access points were
active in the test environment being serviced by this controller, with 1 mobile device, 5 asset tags, 22
rogue access points, and 7 rogue clients present. The trace in Appendix A—Polling Traffic 2700 <->
4400 WLAN Controller, page 109 shows one 60-second polling window where all polling was
completed in just under 11 seconds via a 100BaseT switched LAN. As can be seen, 39 SNMP “Get”
commands were issued to the WLAN controller with 39 responses returned to the location appliance.
Careful analysis indicates that for this polling cycle, 10,243 bytes were transmitted from the location
appliance to the 4402 via 39 frames, while 153,083 bytes in 117 frames (fragmented responses) were
received by the location appliance as responses.
To examine what traffic volume might be in a larger, more active environment, Appendix C—Large Site
Traffic Analysis, page 116, contains an analysis of the traffic flow between the location appliance and
WLAN controllers in a large scale, multi-floor building with a very busy WLAN. The test infrastructure
is 10/100/1000 Ethernet and includes the following:
• Single WCS and Location Appliance
• Two Cisco WLC4400-100 WLAN Controllers (controller #1, controller #2)
• 41 access points (15 on controller #1, 26 on controller #2)
• 245 clients (114 on controller #1, 131 on controller #2)
• 172 asset tags (103 on controller #1, 69 on controller #2)
• 517 rogue access points
• 27 rogue clients (4 on controller #1, 23 on controller #2)
• Location Server running for 14 days
To gain a better understanding of not only the aggregate polling traffic flow but also what can be
expected when enabling each individual polling category in this environment, trace information and
traffic statistics were gathered as follows:
• Location appliance polling asset tags only
• Clients only
• Rogues only
• Statistics only
This was accomplished by disabling all polling on the location appliance except for the specific category
under test. Multiple polling cycles were captured at clearly distinguishable intervals to ensure that valid
data was obtained. The data was then carefully groomed so that the traffic analysis included only the
SNMP commands and responses from a single polling cycle. For comparison, Appendix C—Large Site
Traffic Analysis, page 116, contains an analysis of the aggregate traffic flow between the location
appliance and WLAN controllers when all four polling categories are enabled simultaneously for a
120-second polling interval.
When reviewing the data in Appendix C—Large Site Traffic Analysis, page 116, keep in mind that a
large amount of the traffic occurring between the location appliance and the WLAN controllers is
because of an unusually high number of rogue access points being present. Although this may be