Detecting Microburst Traffic

Context

Microburst detection is used to detect instantaneous (millisecond level) burst traffic received on an interface. This function allows you to view statistics on key indicators of microburst traffic and collect statistics on packet loss. In this way, you can identify potential congestion risks on the network.

A microburst refers to a situation in which a large amount of burst data is received in milliseconds on an interface, so that the instantaneous data rate is tens or hundreds times higher than the average rate or even exceeds the interface bandwidth. When a microburst exceeds the forwarding capability of a switch, the switch buffers the burst data for later transmission. If the switch does not have sufficient buffer space, the excess data is discarded, causing congestion and packet loss. Traditional fault locating methods are complex and difficult. Typically, after packets are lost due to congestion on an outbound interface, outgoing packets are obtained to analyze the traffic trend and identify the characteristics of burst traffic. Maintenance personnel can use the microburst detection function to check whether packet loss is caused by microbursts. Microburst detection helps to identify potential congestion risks before they occur and quickly locate abnormal traffic after congestion occurs.

The following microburst detection modes are supported:

Default mode: Packets are sampled at an interval of 5 ms. In this mode, microburst detection can be enabled on multiple interfaces.
Enhanced mode: Packets are sampled at an interval of 1 ms. In this mode, microburst detection can be enabled on only one interface.

The measurement period of microburst detection is 5 minutes. That is, the key performance indicators of interfaces are collected every 5 minutes and related entries are generated accordingly. The switch can store statistics collected within up to 300 minutes after microburst detection is enabled. The key performance indicators of microburst detection are as follows:

Average rate of burst traffic forwarded to an interface from any other interfaces on the same device
Peak rate of burst traffic forwarded to an interface from any other interfaces on the same device
Number of discarded packets on an interface
Average buffer usage of an interface
Peak buffer usage of an interface
Buffer usage of interface queues when the interface buffer usage reaches the peak value in a measurement period

Only the S5730-HI, S5731-H, S5731-S, S5731S-S, S5731S-H, S5732-H, S6720-HI, S6730-H, S6730S-H, S6730-S, and S6730S-S support microburst detection.

Procedure

Run system-view
The system view is displayed.
Run qos micro-burst detection [ enhanced ] enable slot slot-id
Microburst detection is enabled globally.

By default, microburst detection is disabled on a switch.
Run interface interface-type interface-number
The interface view is displayed.
Run qos micro-burst detection enable
Microburst detection is enabled on the interface.

By default, microburst detection is disabled on an interface.
Run quit
Return to the system view.

Follow-up Procedure

Run the following commands to check statistics on microburst detection:

Run the display qos micro-burst peak-buffer verbose interface interface-type interface-number command to check the peak buffer usage and the buffer usage of queues on an interface.
Run the display qos micro-burst statistics interface interface-type interface-number command to check the key statistics collected after microburst detection is enabled on an interface, including the average rate of burst traffic, maximum rate of burst traffic, number of discarded packets, average buffer usage, maximum buffer usage, and entry generation time.
Run the display qos micro-burst status all [ slot slot-id ] command to check all interfaces enabled with microburst detection and packet loss information on the interfaces.

You can take the following countermeasures to mitigate microbursts:

For problems such as severe bursts, intensive buffer utilization, poor performance on lossy lines, and large delay and jitter in traditional TCP congestion control mechanisms, use common improvement technologies in the industry to minimize the possibility of microbursts.
During network service traffic planning, avoid scenarios with an excessively high convergence ratio (traffic is transmitted from multiple ports to one port), and expand the capacity in a timely manner for ports with severe bursts to eliminate burst bottlenecks.
On switches where congestion occurs, to mitigate network congestion, run the qos burst-mode { enhanced | extreme } command in the interface view to set the burst traffic buffer mode to enhanced.
If the delay is controllable and the buffer is sufficient, run the qos queue queue-index shaping cir cir-value pir pir-value [ cbs cbs-value pbs pbs-value ] command to enable the traffic shaping function on downlink ports of the upstream switch connected to the switch on which congestion occurs. This reduces the instantaneous peak of the traffic and controls the burst. Be aware that this solution will increase the packet forwarding delay.