Hello Ahmed,
This table is a strange way to put things. Let me elaborate a little, and I'll try to explain why I think the table is wrong.
When doing a IPv6-in-IPv4 tunnel, the tunnel may be terminated at the actual end host that is the final recipient of this packet (this assumes that the host is dual-stacked and capable of performing this appropriate tunneling), or it may be terminated at a router that provides the tunneling services for the entire network behind it (in this case, the end hosts in the inside network may be pure IPv6 and only the router is dual-stacked).
If the tunnel is to be terminated at the end host then the IPv6 address must embed the end host's IPv4 address. Recalling the formats of the addresses, it is:
- 6to4: 2002::<16 BIT SUBNET ID>:<64 BIT HOST ID>
- ISATAP: <64 BIT GLOBAL PREFIX>:0000:5EFE:
Now, each of these addresses uniquely carries the end host's IPv4 address and also provides unique IPv6 addressing - for different end hosts, both 6to4 and ISATAP addresses derived from the end host's IPv4 address are unique.
If the tunnel is to be terminated at a tunneling router, the situation becomes different:
- 6to4: 2002::<16 BIT SUBNET ID>:<64 BIT HOST ID>
- ISATAP: <64 BIT GLOBAL PREFIX>:0000:5EFE:
Here, observe that while 6to4 addresses are still unique for each internal host (they differ in the 64 bit host ID), this is not the case for ISATAP addresses. Different end hosts would be assigned identical ISATAP addresses because in this ISATAP address, there is nothing specific to the end host. So that makes the ISATAP tunneling inappropriate for deployments in which a single router is providing tunneling services for its entire internal network. No surprise, after all: the ISATAP was designed from the beginning as a tunneling method for end hosts.
And this is what I do not like about the table you have quoted. From what I explained just now, it should be fairly obvious that the 6to4 tunneling is fine both for end hosts performing the tunneling and for routers providing tunneling services. The ISATAP tunneling, however, is not directly applicable to routers providing tunneling services because it does not result into unique IPv6 addresses for individual end hosts.
It seems like the "Yes" and "No" words should be swapped in the ISATAP column.
Best regards,
Peter
