Understanding Node Groups
Node groups help you combine multiple Node devices into a single virtual entity within the QFabric system to enable redundancy and scalability at the edge of the data center.
Network Node Groups
A set of one or more Node devices that connect to an external network is called a network Node group. The network Node group also relies on two external Routing Engines running on the Director group. These redundant network Node group Routing Engines run the routing protocols required to support the connections from the network Node group to external networks.
When configured, the Node devices within a network Node group and the network Node group Routing Engines work together in tandem as a single entity. By default, network Node group Routing Engines are part of the NW-NG-0 network Node group but no Node devices are included in the group. As a result, you must configure Node devices to be part of a network Node group.
In a QFabric system deployment that requires connectivity to external networks, you can modify the automatically generated network Node group by including its preset name NW-NG-0 in the Node group configuration. Within a network Node group, you can include a minimum of one Node device up to a maximum of eight Node devices. By adding more Node devices to the group, you provide enhanced scalability and redundancy for your network Node group.
The QFabric system creates a single NW-NG-0 network Node group for the default partition. You cannot configure a second network Node group inside the default partition. The remaining Node devices within the default partition are reserved to connect to servers, storage, or other endpoints internal to the QFabric system. These Node devices either can be retained in the automatically generated server Node groups or can be configured as part of a redundant server Node group.
Server Node Groups
A server Node group is a set of one or more Node devices that connect to servers or storage devices. Unlike Node devices that are part of a network Node group and rely on an external Routing Engine, a Node device within a server Node group connects directly to endpoints and implements the Routing Engine functions locally, using the local CPU built into the Node device itself.
There are two different server Node group types:
Autogenerated server Node group—By default, each Node device is placed in its own self-named Node group to connect to servers and storage. No configuration is necessary, but the QFabric system provides no redundancy for this type of Node group.
Redundant server Node group—You can override the default, autogenerated server Node group assignment by manually configuring a redundant server Node group that contains a maximum of two Node devices. You can use a redundant server Node group to provide multihoming services to servers and storage, as well as configure aggregated LAG connections that span the two Node devices.
The Node devices in a redundant server Node group must be of the same type, either two QFX3500 Node devices, two QFX3600 Node devices, or two QFX5100 Node devices. You cannot mix and match different Node device models in the same redundant server Node group.