Re: Shared Locator Address Pool (SLAP) protocol proposal

[Pekka Nikander <pekka.nikander@nomadiclab.com>]

Brian E Carpenter wrote:

> Spencer, I think the point here is that there will be state
> to maintain, and each user of that state will need to access a
> common STorage Information bloCK, i.e. the SLAP STICK.
>
> The kind of shim that NOID suggests could also be a user
> and maintainer of the slapstick.
>
> I'd encourage people to think about this as a potential
> architectural component, in the spirit of the final part of
> the discussion last week.

Personally, I like the SLAP STICK idea very much.  To me,
it is pretty clear that the interface between multi-addressed
layer 3.5 and transport has some serious issues, as discussed
e.g. by Spencer.  Perhaps SLAP, maybe combined with ECM,
could do something to them.

However, I am also worried about potential DoS and other
security issues.  To me, it looks like a bad idea of allowing
all of the upper layer protocols to add or remove addresses
from SLAP.  Updating (soft) address state is probably fine,
but both adding and deleting addresses is potentially dangerous.

Hence, maybe we should consider a new protocol to maintain
the SLAP state, and instead of having upper layer address
update mechanism (eg. SCTP addip), route the upper layer
APIs to the new protocol.

--Pekka Nikander