What are the recommendations for deploying the Aruba Mobile Edge architecture across low-speed and high-latency links using ArubaOS 3.x?
Product and Software: This article applies to all Aruba controllers and ArubaOS 3.1, 3.2, and 3.3.
Deploying APs over Low-Speed Links
Aruba recommends that controllers and APs be connected over a link with a capacity of 1 Mbps or greater, and requires a minimum link speed of 64 Kb/s per tunnel-mode SSID. LANs, many metro area networks, and most broadband DSL or cable connections today provide link speeds greater than 1 Mbps. For low-speed links such as ISDN, T1/E1, and many frame relay WAN connections, specific design guidelines and Aruba configuration are necessary to avoid connectivity problems.
To achieve high reliability and fast failover in the event of a network or controller outage, APs and controllers maintain "heartbeat" or "keepalive" packets between themselves. Failure to receive these heartbeat packets, which are described in detail next, can cause APs to "rebootstrap", which means to go through a process of tunnel re-establishment with the controller. During the bootstrap process, the AP shuts off all radios for approximately 20 ms, and all clients will be required to re-associate.
Two types of heartbeat packets are sent: GRE tunnel keepalives and PAPI keepalives:
GRE Tunnel Keepalives
· Sent once each second.
· Bi-directional: Sent from AP to controller and from controller to AP.
· By default, AP rebootstraps after missing eight consecutive keepalive packets, which can be adjusted by using the AP > AP system profile > bootstrap thresholdparameter.
· By default, the controller will remove an AP's tunnel after 1.5 times the bootstrap-threshold, which defaults to 12 seconds of inactivity.
· Does not apply to bridge mode SSIDs (Remote AP). APs in this mode will not have GRE tunnel to the controller.
· The PAPI-based keepalive messages are sent by the AP every keepalive interval (default 60 in AP > AP system profile).
· Keepalive message are retransmitted up to maximum request retries (default is 10 in the AP system profile). The interval between the first and second attempts is given by request retry interval (default 10 in the AP system profile). With subsequent attempts, the interval grows by a factor of 1.25 up to a maximum of 30 seconds. Intervals of at least 20 seconds will be randomized by +/- 5 seconds.
· After the maximum number of retries, the AP rebootstraps (with default settings, it should take a bit under 4 minutes for this to happen).
· Effectively applies only to bridge-mode SSIDs. For tunnel-mode SSIDs, GRE tunnel keepalive intervals will time out long before PAPI keepalive intervals.
All these parameters can be modified from the GUI:
Configuration > All Profiles > AP > AP system profile