Hash-based load balancing

Hash-based load balancing shares traffic load across LAG ports, while ensuring that packets in the flow are not reordered.

Note

Note

The MPLS options in this section are not supported for SLX 9150 and SLX 9250 devices.

Configuring LAG hashing

To configure symmetric LAG hashing on supported devices, complete the following tasks.

  1. Define where to start picking headers for the key generation, using the lag hash hdr-start command.
    • fwd—Start from the header that is used for the forwarding of the packet (inner header). This is the default option.
    • term—Start from the last terminated header (outer header)—the header after the forwarding header. For switching traffic, as there is no header below the forwarding header, hashing is not visible.
  2. Configure the number of headers to be considered for LAG hashing, using the lag hash hdr-count command. The default value is 1. There can be a maximum of 3 headers—based on the first header selected using the command in the previous step.

The following options provide other LAG configurations to achieve specific tasks:

Configuring header protocols for load-balancing

Select the protocol header type using one of the following commands. By default, all the header parameters are enabled as shown here. If you disable a header, you can then re-enable its parameters one-by-one.

Load balancing mechanism on different traffic types

The following table provides information about load balancing on different traffic types.

Table 1. Load balancing on different traffic types
Traffic type Header field Description
Layer 2/ Layer 3 packet load balancing
  • Ethernet DA, SA, Etype, Vlan-id
  • IPv4/v6 dst IP, src IP
  • L4 Src-Port, Dst-Port
  • Ethernet destination address, source address, ethernet type, VLAN ID load balancing
  • IPv4/v6 destination address, source address load balancing
  • Layer 4 source and destination port-based load balancing
VPLS/ VLL packet load balancing CE to PE router traffic can use the following fields for load-balancing similar to the Layer 2/ Layer 3 traffic)
  • Ethernet DA, SA, Etype, Vlan-id
  • IPv4/v6 dst IP, src IP
  • L4 Src-Port, Dst-Port
PE to CE router traffic can use the following fields for load-balancing
  • Customer (inner) ethernet DA, SA, Etype, Vlan-id
  • Customer (inner) IPv4/v6 dst IP, Ipv4/Ipv6 src IP, protocol
  • Customer (inner) L4 Src-Port, Dst-Port

CE to PE router traffic

  • Ethernet destination address, source address, ethernet type, VLAN ID load balancing
  • IPv4/v6 destination address, source address load balancing
  • Layer 4 source and destination port-based load balancing

PE to CE router traffic

  • Customer ethernet destination and source address, ethernet type, VLAN ID load balancing
  • Customer IPv4/v6 destination address, source address load balancing
  • Customer Layer 4 source and destination port-based load balancing
MPLS LSR load balancing
  • Hashing options support different MPLS transit hashing scenarios
  • The hashing options are mutually exclusive. If one option is enabled, the other option will be disabled.
IP over MPLS traffic going over transit node Extreme supports speculate-mpls option as default which speculates the IPv4/IPv6 header after the MPLS labels and use the fields for hashing. This hashing scenario is handled by the lag hash speculate-mpls enable command in the global mode.
L2VPN (VPLS/VLL) traffic
  • The hashing options are mutually exclusive. If one option is enabled, the other option will be disabled.
L2VPN tagged mode with IPv4 inner payload This scenario is handled using the lag hash speculate-mpls inner-ip-tag command in the global mode. Some sections of the IPv4 source and destination address fields are also used for load-balance hashing.
L2VPN raw mode with IPv4 inner payload This scenario is handled using the lag hash speculate-mpls inner-ip-raw command. Some sections of the IPv4 source and destination address fields are also used for load-balance hashing.
L2VPN tagged mode with IPv6 inner payload This scenario is handled using the lag hash speculate-mpls inner-ipv6-tag command. Some sections of the IPv6 source and destination address fields are also used for load-balance hashing.
L2VPN raw mode with IPV6 inner payload This scenario is handled using the lag hash speculate-mpls inner-ipv6-raw command. Some sections of the IPv6 source and destination address fields are also used for load-balance hashing.