Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
ON THIS PAGE
 

Understanding ECMP Groups

SUMMARY 

Configuring Consistent Load Balancing for ECMP Groups

Per-packet load balancing allows you to spread traffic across multiple equal-cost paths. By default, when a failure occurs in one or more paths, the hashing algorithm recalculates the next hop for all paths, typically resulting in the redistribution of all flows. Consistent load balancing enables you to override this behavior so that only flows for links that are inactive are redirected. All existing active flows are maintained without disruption. In a data center environment, the redistribution of all flows when a link fails potentially results in significant traffic loss or a loss of service to servers whose links remain active. Consistent load balancing maintains all active links and instead remaps only those flows affected by one or more link failures. This feature ensures that flows connected to links that remain active continue uninterrupted.

This feature applies to topologies where members of an equal-cost multipath (ECMP) group are external BGP neighbors in a single-hop BGP session. Consistent load balancing does not apply when you add a new ECMP path or modify an existing path in any way. To add a new path with minimal disruption, define a new ECMP group without modifying the existing paths. In this way, clients can be moved to the new group gradually without terminating existing connections.

  • (On MX Series) Only Modular Port Concentrators (MPCs) are supported.

  • Both IPv4 and IPv6 paths are supported.

  • ECMP groups that are part of a virtual routing and forwarding (VRF) instance or other routing instance are also supported.

  • Multicast traffic is not supported.

  • Aggregated interfaces are supported, but consistent load balancing is not supported among members of the link aggregation (LAG) bundle. Traffic from active members of the LAG bundle might be moved to another active member when one or more member links fail. Flows are rehashed when one or more LAG member links fail.

  • We strongly recommend that you apply consistent load balancing to no more than a maximum of 1,000 IP prefixes per router or switch.

  • Layer 3 adjacency over integrated routing and bridging (IRB) interfaces is supported.

You can configure the BGP add-path feature to enable replacement of a failed path with a new active path when one or more paths in the ECMP group fail. Configuring replacement of failed paths ensures that traffic flow on the failed paths only are redirected. Traffic flow on active paths will remain unaltered.

Note:

  • When you configure consistent load balancing on generic routing encapsulation (GRE) tunnel interfaces, you must specify the inet address of the far end GRE interface so that the Layer 3 adjacencies over the GRE tunnel interfaces are installed correctly in the forwarding table. However, ECMP fast reroute (FRR) over GRE tunnel interfaces is not supported during consistent load balancing. You can specify the destination address on the router configured with consistent load balancing at the [edit interfaces interface name unit unit name family inet address address] hierarchy level. For example:

    For more information on generic routing encapsulation see Configuring Generic Routing Encapsulation Tunneling.

  • Consistent load balancing does not support BGP multihop for EBGP neighbors. Therefore, do not enable the multihop option on devices configured with consistent load balancing.

To configure consistent load balancing for ECMP groups:

  1. Configure BGP and enable the BGP group of external peers to use multiple paths.
  2. Create a routing policy to match incoming routes to one or more destination prefixes.
  3. Apply consistent load balancing to the routing policy so that only traffic flows to one or more destination prefixes that experience a link failure are redirected to an active link.
  4. Create a separate routing policy and enable per-packet load balancing.
    Note:

    You must configure and apply a per-packet load-balancing policy to install all routes in the forwarding table.

  5. Apply the routing policy for consistent load balancing to the BGP group of external peers.
    Note:

    Consistent load balancing can be applied only to BGP external peers. This policy cannot be applied globally.

  6. (Optional) Enable bidirectional forwarding detection (BFD) for each external BGP neighbor.
    Note:

    This step shows the minimum BFD configuration required. You can configure additional options for BFD.

  7. Apply the per-prefix load-balancing policy globally to install all next-hop routes in the forwarding table.
  8. (Optional) Enable fast reroute for ECMP routes.
  9. Verify the status of one or more ECMP routes for which you enabled consistent load balancing.

    The output of the command displays the following flag when consistent load balancing is enabled:State: <Active Ext LoadBalConsistentHash>

Understanding Consistent Load Balancing Through Resilient Hashing on ECMP Groups

You can use consistent load balancing to minimize flow remapping in an equal-cost multipath (ECMP) group.

By default, when there are multiple equal-cost paths to the same destination for the active route, Junos OS uses a hash algorithm to choose one of the next-hop addresses to install in the forwarding table. Whenever the set of next hops for a destination changes in any way, Junos OS rechooses the next-hop address by using the hash algorithm.

You can configure consistent load balancing on the switch to prevent the reordering of all flows to active paths in an ECMP group when one or more next-hop paths fail. Only flows for paths that are inactive are redirected to another active next-hop path. Flows mapped to servers that remain active are maintained.

This feature applies only to external BGP peers.