Comparison Between Stacking and M-LAG

Reliability

Medium

• Control planes are centralized, and faults may be spread on member devices.
• High device-level, card-level, and link-level reliability.

Higher

• Control planes are independent and fault domains are isolated.
• High device-level, card-level, and link-level reliability.

Configuration complexity

Simple: Member devices are logically regarded as one device.

Medium: Two devices need to be configured independently.

Cost

Medium: Stack cables need to be deployed.

Medium: A peer-link needs to be deployed.

Performance

Medium: The master device's control plane needs to control forwarding planes of all stack members, which increases the CPU load.

High: Member switches forward packets independently. The CPU load remains unchanged.

Upgrade complexity

High: Fast stack upgrade reduces service interruption time but increases the upgrade operation time and upgrade risks.

Low: The upgrade is performed by restarting the device, which is simple and of low risk.

Duration of service interruption caused by software upgrade

Long: In typical networking, the service interruption time upon fast stack upgrade is about 20 seconds to 1 minute, which is closely related to the service volume.

Short: The upgrade interrupts traffic in seconds.

Network design

Simple: There is a single node in a stack logically.

Complex: There are two nodes in an M-LAG logically.

Application scenario

• There is no special requirement on the service interruption time during software upgrade.
• Simple maintenance is required.

• The requirement on the service interruption time during software upgrade is high.
• The requirement on reliability is high.
• The increased maintenance complexity can be accepted.