Redundant network design
Figure 24: SMLT scaling
For more information about the SMLT triangle, square, and full-mesh designs, see Avaya Ethernet
Routing Switch 8800/8600 Configuration — Link Aggregation, MLT, and SMLT, NN46205-518.
For more information about SMLT, see the Internet Draft draft-lapuh-network-smlt-06.txt available at
www.ietf.org.
SMLT full-mesh recommendations with OSPF
In a full-mesh SMLT configuration between two clusters running OSPF (typically an RSMLT
configuration), Avaya recommends that you place the MLT ports that form the square leg of the
mesh (rather than the cross connect) on lower numbered slots/ports. This configuration is
recommended because CP-generated traffic is always sent out on the lower numbered MLT ports
when active. This configuration keeps some OSPF adjacencies up in case the IST on one cluster
fails. Without such a configuration, a booted switch in the scenario where the IST is also down can
lose complete OSPF adjacency to both switches in the other cluster and therefore become isolated.
Routed SMLT
In many cases, core network convergence time depends on the length of time a routing protocol
requires to successfully converge. Depending on the specific routing protocol, this convergence time
can cause network interruptions ranging from seconds to minutes.
Routed Split MultiLink Trunking (RSMLT) allows rapid failover for core topologies by providing an
active-active router concept to core SMLT networks. RSMLT is supported on SMLT triangles,
squares, and SMLT full-mesh topologies that have routing enabled on the core VLANs. RSMLT
provides redundancy as well: if a core router fails, RSMLT provides packet forwarding, which
eliminates dropped packets during convergence.
June 2016
Planning and Engineering — Network Design
Comments on this document? infodev@avaya.com
86