Understanding Fabric Status Updates, vPorts and Keepalives


Instead of transactions, which are reserved for create, delete, modify, add or remove commands, Netvisor ONE uses status update messages to distribute information across the fabric.


Unlike transactions, Netvisor ONE does not strictly guarantee the status update data to be synchronized across the fabric.


Netvisor ONE sends the following states from the local node to the other nodes in the fabric:


  • Node State
  • Port State
  • VLAN State
  • Owned vPort State
  • Layer 3 Entry State

Netvisor ONE uses vPort (or virtual port) as Layer 2 entries managed by the software and associated to ports where a node performs MAC address learning. vPorts contain information such as the MAC and IP address of a host, the VLAN and the connected port, the state, and other parameters.


In a fabric a node only sends status updates for vPorts that “is part of the fabric, which means that it sends updates only for the states of the hosts directly connected to that node. Netvisor ONE uses the term 'owned vPorts' to represent directly connected hosts.

You can display this information about directly connected hosts using the command, vport-show.


For example:


CLI (network-admin@switch) > vport-show


mac                        vlan        ip                ports        state        migrate

-----------------        ----        --------        -----        -----        ------

00:00:4c:06:91:f8        514        10.81.114.21        61        active          9

                                               

In addition, Netvisor ONE sends status updates whenever a state changes on the fabric. For example, if a port goes down, or you create a new VLAN, the node with the port or VLAN sends a status update about the specific change.

If sending the update message fails, Netvisor ONE attempts to resend it every 250ms.


With multiple nodes configured in the same fabric instance, switches send out special messages, called fabric keepalives, typically every 10 seconds to allow fabric peers to keep track of the online state of other nodes in the same fabric instance.


If a fabric keep-alive message is not received from a specific node for a period of time (30 s) equal to 3 times the transmission interval (10 s), the suddenly-gone-silent node’s state is marked as offline by its neighbors. This inter-device connectivity/status check is performed with every other node within the same fabric instance.