[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: CELP (was RE:)
>> I still operate under the possibly-false view of an Internet that is
>> mostly reachable. Perhaps partitioning is already too much of a
>> for us to ignore this type of fine-grained mechanism?
> This is very much the world of *today*.
If by "partition" dave meant "topology parition", it is very minimal.
The so-called dark or murky address space are well studied by arbor
network group (labovitz et all.) Dark address space is reachable
through one provider, but not accessible through some other providers.
It is due to broad band customers. The murky address space is due to
misconfiguration or use of 1918 address space or reserved ones. I did
some measurements last year and found it to be very minimal(which
you can find in caida ISMA 02 workshop.)
anyway, i am still puzzled by the whole partition argument. In 9/11
case, operators did reroute traffic although many outages were observed
in the NY area ( verio's junction got busted?) But, overall reachability
was fine. OTOH, new sites had flash crowd problem. Hence, it is very
essential for enterprise services to multihome and load balance/fail
over across geographically distributed data centers.
content distribution networks presents a legitimate partition case
where certain prefix can be reached by a subset of nodes and not by
or, you mean black holes, port blocking etc contribute to partitions?
As a side note, how is HIP going to allow port blocking? Will it avoid
worm attacks by its puzzle mechanism. I don't think it is possible (but
it can reduce its spawning speed.)