168 UtilityNonstop Forwarding on a Switch StackNetworking devices, such as the PowerConnect 6200 Series switches, are often described in terms ofthree semi-independent functions called the forwarding plane, the control plane, and the managementplane. The forwarding plane forwards data packets and is implemented in hardware. The control plane isthe set of protocols that determine how the forwarding plane should forward packets, deciding whichdata packets are allowed to be forwarded and where they should go. Application software on themanagement unit acts as the control plane. The management plane is application software running onthe management unit that provides interfaces allowing a network administrator to configure the device.The Nonstop Forwarding (NSF) feature allows the forwarding plane of stack units to continue to forwardpackets while the control and management planes restart as a result of a power failure, hardware failure,or software fault on the stack management unit. This type of operation is called nonstop forwarding.When the management unit fails, only the switch ASICs on the management unit need to be restarted.To prevent adjacent networking devices from rerouting traffic around the restarting device, the NSFfeature uses the following three techniques:1 A protocol can distribute a part of its control plane to stack units so that the protocol can give theappearance that it is still functional during the restart.2 A protocol may enlist the cooperation of its neighbors through a technique known as graceful restart.3 A protocol may simply restart after the failover if neighbors react slowly enough that they will notnormally detect the outage.Initiating a FailoverThe NSF feature allows you to initiate a failover, which results in a warm restart of the master unit in thestack. Initiating a failover reloads the management unit, triggering the backup unit to take over. Beforethe failover, the management unit pushes application data and other important information to thebackup unit. Although the handoff is controlled and causes minimal network disruption, someapplication state is lost, such as pending timers and other pending internal events.CheckpointingSwitch applications (features) that build up a list of data such as neighbors or clients can significantlyimprove their restart behavior by remembering this data across a warm restart. This data can either bestored persistently, as DHCP server and DHCP snooping store their bindings database, or themanagement unit can checkpoint this data directly to the backup unit. Persistent storage allows anapplication on a standalone unit to retain its data across a restart, but since the amount of storage islimited, persistent storage is not always practical.The NSF checkpoint service allows the management unit to communicate certain data to the backupunit in the stack. When the stack selects a backup unit, the checkpoint service notifies applications tostart a complete checkpoint. After the initial checkpoint is done, applications checkpoint changes totheir data.