Hi Steinar,
On Aug 11, 2009, at 4:11 AM, [email protected] wrote:
The dataset analyzed is not relevant to today's networking
connectivity or technologies. Looking very quickly at a small set
of
data I have access to (servers serving web content to the internet
users):
32,945,810,591 packets received, 0 dropped due to bad checksum (ip
header checksum)
1,004,728,008 datagrams received, 0 bad checksum, 15886 with no
checksum (udp datagram stats)
Checking six of our name servers here (some pure recursive, some pure
authoritative):
- 11,125,766,153 IP packets received, with 0 bad header checksums
- 9,311,954,730 UDP datagrams received, with 1,300,228 bad checksums
Thus IP header checksum really looks like it is very close to zero,
probably because any IP packets with bad header checksums would be
dropped by the closest router.
If LISP uses UDP checksums of zero with IPv6, packets with IP header
corruption will not be dropped by the next router, they will be
forwarded all the way to the ETR (or, in the case of destination
address corruption) all the way to the wrong ETR or a non-LISP end-node.
However, the UDP bad checksum rate is very definitely not zero. I
get a rate of 1.39e-4 for my dataset.
This is potentially an argument for using UDP (or UDP-Lite) checksums
in LISP for both IPv4 and IPv6, so that the corruption in the outer
UDP header and the LISP header will be detected before these packets
are processed on the remote end.
Margaret
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
