To initiate a layer 2 roam, the station sends an Associate (or Reassociate) Request to the new AP. It may Disassociate...
from the old AP, or the old AP may notice the station is no longer there. Depending upon the product, there may be some non-standard interaction between the old AP and the new AP to buffer and relay frames destined for the station during handoff. After the station successfully roams, LAN traffic for the station must be relayed through the new AP. MAC address tables and ARP caches on other devices in the same LAN are automatically updated to reflect this, although there may be some frame loss during transition.
Some organizations put many APs in the same Virtual LAN (VLAN) to enable layer 2 roaming over larger areas where the APs can't be on the same physical LAN segment. However, VLANs only scale so far, which brings us to layer 3 roaming.
Wireless stations roam at layer 3 when they disconnect from one AP and reconnect to another AP in a different IP subnet. After layer 2 roaming completes, layer 3 roaming occurs if the station obtains a new IP address -- for example, using DHCP when wireless interface status changes from up to down to up again. Layer 3 traffic re-routing requires more than updating MAC address tables and ARP caches. Network layer devices (routers and L3 switches) must somehow be told to forward IP packets to the station's new subnet. Without some kind of subnet mobility extension, applications will timeout trying to reach the station's old IP and must be reconnect with the station's new IP.
There are many ways of supporting subnet mobility. For example, the Mobile IP standard uses a Home Agent (HA) to forward IP packets to a Foreign Agent (FA) in the node's new subnet. The HA and FA advertise themselves using the ICMP Router Discovery Protocol (IRDP). When a Mobile IP-enabled station roams to a new subnet, it must discover and register itself with a nearby FA. The FA forwards that request to that station's original HA. If the request is accepted, a tunnel is established between the HA and FA to relay incoming packets sent to the node's original IP address. Outbound packets are routed back through the tunnel from the FA to HA, and then on to their destination. Mobile IP preserves subnet connectivity for roaming stations, but results in sub-optimal routing and longer roaming delay. Some vendors have developed other proprietary layer 3 roaming methods to address these concerns.
Today, layer 3 roaming is far from ubiquitous; subnet mobility methods vary widely and are typically found only in enterprise products like WLAN switches and Mobile VPN servers. Basic layer 2 roaming is supported by all WLAN products, but AP-to-AP extensions to make roaming faster or reduce loss work only in homogenous WLANs.
Dig Deeper on Network management and monitoring
Related Q&A from Lisa Phifer
As the remote workforce increases, network managers and users might opt to set up two concurrent VPN connections from the same remote device. But ... Continue Reading
Is there a difference between a wireless access point vs. a router? Yes -- while the two wireless devices are related, they meet different needs in a... Continue Reading
Learn the differences between site-to-site VPNs vs. remote-access VPNs and find out about the protocols, benefits and the data security methods used ... Continue Reading