How does GVRP work?

FAQ_User — Tue, 26 Nov 2013 18:53:00 GMT

Article ID: 5237

Protocols/Features
GVRP

Solution
Some background about GVRP:

GVRP (Generic VLAN Registration Protocol) actions may be viewed as dynamic VLAN creation accompanied by dynamic, tagged egress permissions for the ingress port through which the given VLAN was learned.
Though the resulting functionality could be similar to that seen on a 802.1Q Trunk, ports are not automatically put into "802.1Q Trunk mode" for those products supporting such a mode.
GVRP starts with knowledge of one or more VLANs egressing ethernet-linked ports, then advertises these VLAN IDs to neighboring switches, propagating dynamic 802.1Q VLAN registration packets between GVRP-aware ports. Upon receipt, any VLAN IDs not already known to the receiving switch are created dynamically, and enabled for tagged egress on the receiving port only.
Note: On the SecureStack, all local VLANs are initially advertised - not just those with at least one active port egress (11670).
All GVRP-aware switches perform this same function, while also relaying the GVRP advertisements of others. Ultimately, all GVRP-aware switches should contain much the same VLANs -- some Static (manually configured), and some Dynamic (created via GVRP).
Functionally, it is then possible to configure, onto perimeter switches on opposite ends of the network, a user port to egress a new VLAN - and GVRP will dynamically create a "tunnel" to propagate that VLAN, end-to-end. The intermediate switches need not have ports egressing these VLANs, in order for the GVRP relay to function.
When all active egress for a VLAN has been removed from a switch, its direct advertisement of that VLAN will cease, and any GVRP-learned knowledge of that VLAN will time out unless reinforced by other switches with direct knowledge of the same VLAN.
When using GVRP over a 802.1Q Trunk port, be aware that GVRP itself is propagated without VLAN tags (and GVRP packets are themselves not associated with - or assigned to - a VLAN). Thus, for it to work, the receiving port must not be configured to filter untagged frames. Typically, a 802.1Q Trunk port does in fact filter untagged frames by default.
It is possible to use only a single switch as a Static VLAN "point source", rather than using two or more perimeter switches for this purpose; if the remaining switches are SmartSwitch 2000/6000 2nd/3rd Generation, with all Inter-Switch Links configured as 802.1Q Trunks but permitted to accept both tagged and untagged frames. Whenever a VLAN is learned dynamically on these switches via incoming untagged GVRP advertisements, the VLAN automatically gets tagged egress on all the switches' 802.1Q Trunks - a very nice flexiblity.
Even on devices which support tag insertion/removal on the ATM link, all the VLANs which traverse a given ELAN (this is not a problem for PVCs) must share a single FID. This is due to the fact that ATM LE_ARPs do not contain VLAN information, ever.

GVRP may be classed with GMRP (Generic Multicast Regstration Protocol) as both using the Generic Attribute Registration Protocol (GARP).

See also: 5462.

topic How does GVRP work? in FAQs

How does GVRP work?