Was just wondering if anyone knows if there would be any issues with
using the SSLEngine with a UDP non-blocking channel. I dont really
have anything to test it with lost UDP packets or having the packets
arrive out of order, would this cause problems? are there any utilities
to simulate this connectionless-aspect of UDP?
Thanks,
Chris
Chris Uppal - 06 Nov 2006 13:20 GMT
> Was just wondering if anyone knows if there would be any issues with
> using the SSLEngine with a UDP non-blocking channel. I dont really
> have anything to test it with lost UDP packets or having the packets
> arrive out of order, would this cause problems? are there any utilities
> to simulate this connectionless-aspect of UDP?
You could try using a TCP session (as a test) and deleting / rearanging blocks
of data randomly. Perhaps implemented in some sort of intermediary program.
Personally, I can't see how the encryption (inherently stateful) could possibly
survive in the face of missing or rearanged blocks, but I'm not an SSL expert.
-- chris
EJP - 07 Nov 2006 08:01 GMT
> Was just wondering if anyone knows if there would be any issues with
> using the SSLEngine with a UDP non-blocking channel. I dont really
> have anything to test it with lost UDP packets or having the packets
> arrive out of order, would this cause problems? are there any utilities
> to simulate this connectionless-aspect of UDP?
As Chris suspected, the SSL encryption algorithms won't survive a
dropped packet and indeed under a strict interpretation of RFC2246 will
probably force you to close the connection. You can't implement SSL over
UDP, you can only implement SSL over a reliable transport. So first you
need a UDP reliability layer.