Categories
CCNP Enterprise Core (350-401) Cisco Datalink Layer

Load Balancing Traffic with Etherchannel

Traffic that flows across a port-channel is not forwarded out member links in a round robin basis. A hash is calculated and packets are consistently forwarded across a link based on that calculated hash.

The load balancing hash is a system wide configuration that uses a global command: port-channel load-balance X where X is the hash from the following selections:

Hash algorithmDescription
dst-ipBalances based on destination IP and MAC address only
dst-macBalances based on destination MAC address only
dst-mixed-ip-portBalances based on destination IP and MAC address and TCP/UDP port
dst-portBalances based on destination port only
src-dst-ipBalances based on source IP and destination IP address and MAC addresses
src-dst-ip-onlyBalances based on source IP and destination IP address without MAC address information
src-dst-macBalances based on source and destination MAC addresses
src-dst-mixed-ip-portBalances based on source and destination IP addresses along with source and destination TCP/UDP ports
src-dst-portBalances based on source and destination TCP/UDP ports only
src-ipBalances based on source IP address only
src-macBalances based on source MAC address only
src-mixed-ip-portBalances based on source IP address and TCP/UDP port
src-portBalances on source port only
Types of hash algorithms that may be available on a Cisco switch

Changing the hash may be beneficial to systems that see very high utilisation on one or two member links in a port channel.

Hashing is a binary function; it is important to note that load balancing works best in situations where the number of links are in powers of two. A two or four link etherchannel will load balance better than a three or five link etherchannel.

The command show etherchannel load-balance will display how the switch will load balance its network traffic.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.