Quanta Cloud Technology QuantaMesh QNOS5, Configuration Manual

Page: 75 / 256

75
3.5.1. Overview
In a typical layer-2 network, the Spanning Tree Protocol (STP) is deployed to avoid packet storms due to
loops in the network. To perform this function, STP sets ports into either a forwarding state or a blocking
state. Ports in the blocking state do not carry traffic. In the case of a topology change, STP reconverges to a
new loop-free network and updates the port states. STP is relatively successful mitigating packet storms in
the network, but redundant links in the network are blocked from carrying traffic by the spanning tree
protocol.
In some network deployments, redundant links between two switches are bundled together in a Port-
channel and appear as a single link in the spanning tree topology. The advantage is that all Port-channel
member links can be in the forwarding state and a link failure can be recovered in milliseconds. This allows the
bandwidth on the redundant links to be utilized. However, Port-channels are limited to connecting multiple
links between two partner switches, which leaves the switch as a single point of failure in the topology.
QNOS MLAG extends the Port-channel bandwidth advantage across multiple QNOS switches connected to a
Port-channel partner device. The Port-channel partner device is oblivious to the fact that it is connected
over a Port-channel to two peer QNOS switches; instead, the two switches appear as a single switch to the
partner with a single MAC address. All links can carry data traffic across a physically diverse topology and in
the case of a link or switch failure, traffic can continue to flow with minimal disruption.
3.5.2. Deployment Scenarios
MLAG is intended to support higher bandwidth utilization in scenarios where a redundant layer-2 network is
desired. In such scenarios the effects of STP on link utilization are profound. Large percentages of links do not
carry data because they are blocked and only a single path through the network carries traffic.
Figure 3-5: STP Blocking
MLAG reduces some of the bandwidth shortcomings of STP in a layer-2 network. It provides a reduced
convergence period when a port-channel link goes down and provides more bandwidth because all links can
forward traffic. In the figure below, if SW1 and SW2 form a MLAG with SW3 and SW4, none of the links are
blocked, which means traffic can flow over both links from SW4 through to SW1 and SW2 over both links
from SW1 and SW2 to SW3.