ADC

Distributing traffic across cluster nodes

After you have created the NetScaler cluster and performed the required configurations, you must deploy Equal Cost Multiple Path (ECMP) or cluster Link Aggregation (LA) on the client data plane (for client traffic) or server data plane (for server traffic). These mechanisms distribute external traffic across the cluster nodes.

Policy-based backplane steering

The policy-based backplane steering (PBS) is a mechanism in cluster deployment, which steers the traffic across cluster nodes based on the hash method defined for the flow. The flow is defined by a combination of L2 and L3 parameters similar to the Access Control List (ACL).

The PBS supports both IPv4 and IPv6 traffic. In the case of IPv6 deployments, the steering supports an extra option [dfdprefix <positive_integer>]. It provides the flexibility to choose the same flow processor for the same IP prefix. The prefix option is supported for source IP or destination IP hash methods only.

Note

If the PBS mechanism is not used to steer the traffic, the traffic is steered through the default method.

To configure the new ACL attributes, at the CLI, type the following commands:

CLI commands for IPv4

  • add ns acl <aclname> <aclaction> [-type (classic | dfd)] [-dfdhash <dfdhash>]
  • set ns acl <aclname> <aclaction> [-dfdhash <dfdhash>
  • show ns acl [<aclname>][-type (classic | DFD)]
  • apply ns acls [-type (classic | DFD)]
  • clear ns acls [-type (classic | DFD)]
  • renumber ns acls [-type (classic | DFD)]

CLI commands for IPv6

  • add ns acl6 <acl6name> <acl6action> [-type (classic | dfd)][-dfdhash <dfdhash>][-dfdprefix <positive_interger>]
  • set ns acl6 <acl6name> <acl6action> [-dfdhash <dfdhash>][-dfdprefix <positive_interger>]
  • show ns acl6 [<acl6name>][-type (classic | DFD)]
  • apply ns acls6 [-type (classic | DFD)]
  • clear ns acls6 [-type (classic | DFD)]
  • renumber ns acls6 [-type (classic | DFD)]

The following are the different types of hash methods that you can specify to steer the packet to the flow processor:

  • SIP-SPORT-DIP-DPORT
  • SIP
  • DIP
  • SIP-DIP
  • SIP-SPORT

Limitations

  1. The traffic flow distribution across the cluster nodes is not ensured, as the flow processor is decided by the admin configured rules.
  2. L2 mode is not supported.
  3. The node groups and striped SNIPs are not supported, as there is no deployment scenarios.
  4. MPTCP is not supported.
  5. Support only for TCP, UDP, and ICMP traffic.
  6. Cluster over L3 mode is not supported.
  7. Process local at service level is not supported.
Distributing traffic across cluster nodes