Netgear XCM8810 - 8800 SERIES 10-SLOT CHASSIS SWITCH ユーザーズマニュアル

Understanding Hitless Failover Support

The term hitless failover has slightly different meanings on a modular chassis. On a modular 
chassis, MSMs/MMs do not directly control customer ports; such ports are directly controlled 
by separate processors. When a modular chassis MSM/MM failover occurs, all of the ports in 
the chassis are under the control of separate processors which can communicate with the 
backup MSM/MM, so all ports continue to function.

As described in the section 

on page 64, if you install two 

MSMs/MMs (nodes) in a chassis, one assumes the role of primary and the other assumes 
the role of backup. The primary node provides all of the switch management functions 
including bringing up and programming the I/O modules, running the bridging and routing 
protocols, and configuring the switch. The primary node also synchronizes the backup node 
in case it needs to take over the management functions if the primary node fails. 

The configuration is one of the most important pieces of information checkpointed to the 
backup node. Each component of the system needs to checkpoint whatever runtime data is 
necessary to allow the backup node to take over as the primary node if a failover occurs, 
including the protocols and the hardware dependent layers. For more information about 
checkpointing data and relaying configuration information, see 

on page 66. 

Not all protocols support hitless failover; see 

 for a detailed list of protocols and their 

support. Layer 3 forwarding tables are maintained for pre-existing flows, but subsequent 
behavior depends on the routing protocols used. Static Layer 3 configurations and routes are 
hitless. You must configure OSPF graceful restart for OSPF routes to be maintained, and you 
must configure BGP graceful restart for BGP routes to be maintained. For more information 
about OSPF, see 

, and for more information about BGP, see 

. For routing protocols that do not support hitless failover, the new primary node 

removes and re-adds the routes. 

Protocol Support for Hitless Failover

 summarizes the protocol support for hitless failover. Unless otherwise noted, the 

behavior is the same for all modular switches. 

If a protocol indicates support for hitless failover, additional information is also available in 
that particular chapter. For example, for information about network login support of hitless 
failover, see 

MASTER

In the primary (master) state, the node is responsible for all switch management functions.

STANDBY

In the standby state, leader election occurs—the primary and backup nodes are elected. The 
priority of the node is only significant in the standby state.