Implementing a Highly Available Network Implementing High Availability Implementing a Highly Available Network
Layer 2 Distributed VLANs on Access Switches Not a recommended design — slow convergence Use only if Layer 2 VLAN spanning flexibility is required Requires STP convergence for uplink failure and recovery More complex because the STP root and HSRP should match May be required for WLAN (standalone APs)
Layer 2 Local VLANs on Access Switches Recommended design, tried and true VLANs present on one access switch only Does not require STP convergence for uplink failure recovery Requires a distribution-to-distribution link for route summarization Can map Layer 2 VLAN numbers to Layer 3 subnets for ease of use and management
Layer 3 Access-to-Distribution Interconnection Best option for fast convergence, and easy to implement Uses equal-cost Layer 3 load balancing on all links Does not require STP for convergence Does not require HSRP (FHRP) configuration Does not support VLAN spanning distribution switches
Daisy-Chaining Access Layer Switches Primary and secondary HSRP is active after failure. Outbound traffic is sent from both HSRP instances.
Daisy-Chaining Access Switch Issues Potential for black holes if no alternative path is provided
StackWise Technology Access Switches StackWise Technology eliminates the daisy-chain issue: Loopback links are not required. A Layer 2 link in the distribution is not required. StackWise switch provides redundancy. Uplinks can be on different switches within stack. Modular chassis–based switches can also eliminate the daisy-chain issue.
Avoiding Too Little Redundancy Looped figure-8 topology for VLANs spanning access switches Blocking on uplink from Access-b. Initially forwarding traffic from both access switches
Impact of Uplink Failure Traffic from Access A is dropped until HSRP goes active on Distribution B. Blocking link on Access B takes 50 seconds to move to forwarding. After STP converges, an HSRP preempt causes another transition. Access B is used as transit for Access A traffic.
Summary Achieving redundancy can be achieved at Layer 2 and at Layer 3, by providing additional redundant paths between devices. StackWise Technology can be used to join several physical switches into one virtual switch. Redundancy is a balance between too much redundancy, which adds complexity to the network structure, and too little redundancy, which creates single points of failure. When uplinks fail, convergence path as well as convergence time must be taken into account to evaluate the impact of the failure on the network infrastructure.