L2VPNs - Benefits of / differences between differen port encapsulations
Im aware that a customer using a L2VPN may send traffic over to a PE which is encapsulated as Null port, 802.1q port or Q-in-Q port type.
I think Null encapsulation translates as raw mode / port based VLAN (?) and requires service provider to send customer VLAN info across its network unchanged, in this sense, customer VLAN info must match at both ends.
I think 802.1q/Q-in-Q encpasulation tranlsates as VLAN based VLAN, where a service is mapped to a VLAN (service delimiting) but customer VLAN info is dropped at ingress and provider tags (PVLAN or an MPLS service label) are used inside the network to switch customer frames to correct vlan - cust VLANs can be different at both ends in this sense, it doesnt matter.
What is the beneft of either mode over the other? How important is the encpasulation on the PE port? Can it be anything and still work? Are there combinations that work and some that dont depending on what the customer sends and encaps at the far end - do they have to match?
Im a little confused, im trying to build my knowledge of troubleshooting L2VPNS - many of the faults ive seen involve incorrect port encapsulations for a given customer service (im coming from an Alcatel / Nokia services background but I hope the terms are interchangeable as I'm about to begin on Juniper L2VPN networks)
Hope I explained this the best I could, i know in certain situations the port encpasulations are set up incorrectly for the situation given - i just need this clearing up, when would it not work and why?