Layer 3 Port Adoption

[ToM]

Hi,

In Enterprise WLAN Design Guide, Volume 1 November 2008 it is said that 'WISPE requires a minimum bandwidth of 10 Mbps between the switch and the WiAP to operate without substantial degradation in roaming quality' in paragraph 6.3.4.1.

Why is that? What are the consequencies if the link between the AP and the switch is much less, for example just 1 Mbps. If there is no need to use the full capacity of 802.11g network.

My second question concerns 'Avg Retries per pkt' in switch management GUI. What does that mean. For example if the 'Avg Retries per pkt' in the graph varies between 15 and 35 (or even more). Is there something seriously wrong with the configuration? Or is that result of insufficient bandwith between the AP and the switch.

---

Wireless switch used is RFS7000 and APs are AP-300.

[Michael McNamara]

Hi ToM and welcome to the forums!

In Enterprise WLAN Design Guide, Volume 1 November 2008 it is said that 'WISPE requires a minimum bandwidth of 10 Mbps between the switch and the WiAP to operate without substantial degradation in roaming quality' in paragraph 6.3.4.1.

Why is that? What are the consequencies if the link between the AP and the switch is much less, for example just 1 Mbps. If there is no need to use the full capacity of 802.11g network.

There's a fair amount of overhead with a Layer 3 adopted AP (remember it's essentially tunneling the wireless traffic over an IP connection) and every packet needs to go back to the Wireless LAN switch. In the end you'll obviously need to test your specific situation. What Motorola is trying to say is that unless you have a 10Mbps Full Duplex link between the AP and the Wireless LAN switch you'll likely have throughput and performance issues. One of the issues will be the roaming between APs and the timing involved now that the link between the APs and the Wireless LAN switch is less than expected.

My second question concerns 'Avg Retries per pkt' in switch management GUI. What does that mean. For example if the 'Avg Retries per pkt' in the graph varies between 15 and 35 (or even more). Is there something seriously wrong with the configuration? Or is that result of insufficient bandwith between the AP and the switch.

I believe this is an RF radio statistic. A high value could indicate a lot of RF interference in your physical space which is causing an abnormally high number of retries.

Cheers!

[ToM]

Thanks Michael for the quick response!

I have done some packet capturing with the Ethereal that I found on your site.

The protocol stack goes like this:

DATA (In this case encrypted, Telnet over TCP)
IEEE 802.11
CAPWAP
UDP
IP
Ethernet

Third from the bottom is UDP and as everyone knows that it is connectionless and unreliable protocol. If the used layer 3 link (xDSL) is congested (due to other traffic on the link) and router drops packets will the CAPWAP protocol make sure that every packet is goint to be transferred?

I found that there is Fragment-Id and Fragment-Offset fields in the CAPWAP.

Edit: CAPWAP is not proprietary, but defined in rfc5415. I may have to study that a little.

[Michael McNamara]

You are correct that CAPWAP is not proprietary but WISP and WISPe are Motorola/Symbol proprietary implementations of CAPWAP. I believe the basis of CAPWAP was WISP and Motorola/Symbol were one of the companies involved in bringing the standard forth.

Cheers!