> >We've looked at this concept from time to time over the years. One issue
you
> >have to deal with is that if the checksum fails, can you trust the fields in
> >the packet you use to identify the sender of the packet. There is risk
>
> I ask a much more basic question: how often do TCP checksums fail?
>
> I rebooted my own Solaris workstation about 3 days ago. It talks a lot
> over the exterior Internet. In that time, it has counted two TCP
> errors and one UDP error, which are presumably checksum
> failures. That's out of about 2 million incoming packets.
I am not sure of the intent of the original sender but I was referring to a
link layer checksum, i.e. CRC. Also, I was just pointing this out as something
to take into consideration. Even though we thought about it from time to time,
we never convinced ourselves that the number of packets received with bad CRCs
was high enough (in our environment) to make having to pass link layer
information up the stack worthwhile. (We found that, in most cases, when
errors occured packets were totally lost, not received with bad CRCs.)
John
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:19 EST