Fast ICMP reply function on S series switches

22

After receiving an ICMP Request packet, an S series switch (a non-S1700 switch) does not send it to the ICMP protocol stack but directly exchanges the source and destination MAC addresses of the packet and responds with an ICMP Relay message. This process is called fast ICMP reply. (The switch can successfully ping another device even if it does not have a return route or an ARP entry.) A switch cannot accurately calculate the delay or jitter of ping packets that require high real-time performance. The reason is that the protocol stack adds the sending and receiving timestamps to ping packets. The packets need to enter queues frequently between the protocol stack and the hardware. Therefore, the real-time performance of timestamps cannot be ensured.
In versions earlier than V100R006, fast ICMP reply is a common task. When a switch processes a large number of services, the reply delay is long due to service switching. In V100R006 and later versions, fast ICMP reply is a super task with a high priority. In most cases, the relay delay is about 1 ms. The actual delay depends on the CPU usage and is shorter than 100 ms.
The icmp-reply fast command enables the fast ICMP reply function.
Note: The fast ICMP reply function is disabled on modular switches running all versions by default.

Other related questions:
Whether the S-series switches support fast forwarding
support

Implementation and application of fast leave of S series switches
For an S series switch, the fast leave function is configured using the igmp-snooping prompt-leave command (for Layer 2 multicast) or the igmp prompt-leave command (for Layer 3 multicast). If this function is not configured, the switch sends a Group-Specific Query message in response to an IGMP Leave message to check whether there are other users in the group specified in the Leave message. After fast leave is configured, the switch directly deletes the interface that receives the IGMP Leave message from the multicast forwarding entry, and does not check whether there are still users in the group. The fast leave function is applicable only to interfaces with only one user attached. If fast leave is configured on an interface connecting to multiple users, multicast traffic of all users under the interface is interrupted when one user switches to another channel.

Whether S series switches support ICMP jitter test instances
All S series switches except S3700 switches support ICMP jitter test instances.

For S series switches, what is the impact of excessive ARP Replies on the CPU
For S series switches: If excessive ARP Reply packets are sent to the CPU, the CPU may be overloaded. Run the display cpu-defend configuration packet-type arp-reply all or the display cpu-defend statistics packet-type arp-reply all command to check whether excessive ARP Reply packets are sent to the CPU. In the display cpu-defend statistics packet-type arp-reply all command output. if the value of the Drop (Bytes) field is large, excessive ARP Reply packets are sent to the CPU. In this case, adjust the CPCAR value for the ARP Reply packet. When an attack occurs, determine the attack source. You can use the packet obtaining function or enable debugging to check the attack source and configure the blacklist to reject the attack source. Note: Improper CPCAR settings may affect services on your network. It is recommended that you contact Huawei engineers before adjusting the CPCAR settings.

Configuring IS-IS fast convergence on S series switches
To configure IS-IS fast convergence on S series switches supporting IS-IS, run the following commands in the IS-IS process: timer spf flash-flood timer lsp-generation

If you have more questions, you can seek help from following ways:
To iKnow To Live Chat
Scroll to top