[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Multicast Keyserver Synchronization
>>>>> "Albert" == Albert Levi <firstname.lastname@example.org> writes:
Albert> Marcel Waldvogel wrote:
>> Instead of using the "push" semantics using mail, our proposal is
>> to use a combined "push-pull" semantics using IP multicast (and
>> designated unicast tunnels for sites not connected to the MBone).
Hm, this sounds like IS-IS or OSPF or similar routing update
distribution protocols. Given that this is a long-studied area in
layer 3, it's probably a good idea to examine that literature. There
are plenty of pitfalls to be avoided, and at this point it is known
how to do so.
>> * Sender: - it tags that packet with a world-wide unique sequence
>> number, consisting of its host ID to make everything unique, and
>> the next number of its private sequence number space - it
>> multicasts the tuple (host ID, sequence number, OpenPGP packet) to
>> the multicast group (together with an identification of the
>> enclosing UID and public key packet.
For example, sequence numbers raise the issue of how you deal with the
bounded number of bits. There's an obvious way to do this which is
wrong (see paper by Eric Rosen et al. from around 1980) and a right
way which is now used in routing protocols.
Albert> where does a server get its worldwide unique sequence number?
Albert> don't we need a central authority here?
No... Marcel said "consisting of its host ID... and the next number
of its private sequence number space". That's a standard solution: to
get globally unique IDs (often known as UUIDs, you'll find them in
Microsoft stuff but the originated in OSF RPC if not earlier) you use
a MAC address or similar number for spatial uniqueness, and a (locally
managed) time stamp for local temporal uniqueness, plus a sequence
number if needed to cope with low resolution clocks. These field
taken together give you what's needed. (By the way, a timestamp may
be better than a sequence number.)