* Angela Dall'Ara
Dear colleagues,
A new RIPE Policy Proposal, 2024-01, "Revised IPv6 PI Assignment Policy" is now available for discussion.
This proposal aims to define End Sites and requirements for “IPv6 PI Assignments” and “Assignments from IPv6 Allocations”, clarifies permitted use cases and introduces IPv6 PI issuance at the Nibble boundary and new principles for aggregation and registration.
There's one thing I just don't quite get with this proposal. Maybe I am being dense, but hopefully someone can clarify it for me. The proposed text in 2.9 makes it clear that separate locations with different routing policies should be considered separate End Sites. Next, the proposed text in 7.1 makes it clear that having multiple End Sites qualifies for an assignment larger than /48. However, it states that this is done «to avoid fragmentation». But doesn't the fact that those separate End Sites are defined to have different routing policies make the fragmentation happen anyway? Reading between the lines it seems that the whole point of this change is to ensure that the separate End Sites can get their own /48 to advertise in a fragmented manner to the DFZ. From a technical standpoint that makes a lot of sense as the sites might have different IP transit providers and you need at least a /48 to get visibility in the global DFZ. That said, doesn't policy already allow for such End Sites with different routing policies to get their own PI /48? If so, isn't this part of the proposal trying to solve a non-existent problem? Another thing I find strange is the reference to layer-2 connectivity in 2.9 and 7.1.1. This seems oddly technology-specific to me. Is it the intention that two End Sites (with different routing policies) interconnected with an L2VPN should be treated differently than two End Sites (with different routing policies) interconnected with an L3VPN? Looks that way, but why? Tore