Dell S4048T-ON, Configuration Manual

Page: 367 / 1146

Hitless Behavior
Hitless is a protocol-based system behavior that makes a stack unit failover on the local system transparent to remote systems. The system
synchronizes protocol information on the Management and Standby stack units such that, in the event of a stack unit failover, it is not
necessary to notify the remote systems of a local state change.
Hitless behavior is defined in the context of a stack unit failover only.
• Only failovers via the CLI are hitless. The system is not hitless in any other scenario.
Hitless protocols are compatible with other hitless and graceful restart protocols. For example, if hitless open shortest path first (OSPF) is
configured over hitless the link aggregation control protocol (LACP) link aggregation groups (LAGs), both features work seamlessly to
deliver a hitless OSPF-LACP result. However, to achieve a hitless end result, if the hitless behavior involves multiple protocols, all protocols
must be hitless. For example, if OSPF is hitless but bidirectional forwarding detection (BFD) is not, OSPF operates hitlessly and BFD flaps
upon an RPM failover.
The following protocols are hitless:
• Link aggregation control protocol.
• Spanning tree protocol. Refer to Configuring Spanning Trees as Hitless .
Graceful Restart
Graceful restart (also known as non-stop forwarding) is a protocol-based mechanism that preserves the forwarding table of the restarting
router and its neighbors for a specified period to minimize the loss of packets. A graceful-restart router does not immediately assume that a
neighbor is permanently down and so does not trigger a topology change. Packet loss is non-zero, but trivial, and so is still called hitless.
Dell Networking OS supports graceful restart for the following protocols:
• Border gateway
• Open shortest path first
• Protocol independent multicast — sparse mode
• Intermediate system to intermediate system
Software Resiliency
During normal operations, Dell Networking OS monitors the health of both hardware and software components in the background to
identify potential failures, even before these failures manifest.
Software Component Health Monitoring
On each of the line cards and the stack unit, there are a number of software components. Dell Networking OS performs a periodic health
check on each of these components by querying the status of a flag, which the corresponding component resets within a specified time.
If any health checks on the stack unit fail, the Dell Networking OS fails over to standby stack unit. If any health checks on a line card fail,
Dell Networking OS resets the card to bring it back to the correct state.
System Health Monitoring
Dell Networking OS also monitors the overall health of the system.
Key parameters such as CPU utilization, free memory, and error counters (for example, CRC failures and packet loss) are measured, and
after exceeding a threshold can be used to initiate recovery mechanism.
High Availability (HA) 367