ebxml-msg — archive

RE: T2 Retry with Delivery Receipt

From: System
To: [email protected]
Date: Thu, 20 Sep 2001 09:18:00 -0000

David, I just took a quick look at the two articles your posting is pointing to. The first one discusses reasons why 0.1% to 0.3% of TCP packets fail the TCP checksum (the error was not detected by IP or the link). Failing the checksum is not a problem. If TCP doesn't retry, ebXML-MSG will. Nothing in the abstract told me that it would discuss cases where a packet passes checksums in the physical link, IP, and TCP although it is in fact in error. The second article appears to be a huge amount of material. I went through the slide titles and glanced at a few. Again, there is no indication that it discusses cases where a packet makes it all the way through the TCP checksum although it contains transmission errors. Unless you can show evidence that packets are slipping through after failing checksums at all three levels, I agree that we should get off this tangent and leave NRR as strictly a content matter which does not relate to reliable messaging. Regards, Marty ************************************************************************************* Martin W. Sachs IBM T. J. Watson Research Center P. O. B. 704 Yorktown Hts, NY 10598 914-784-7287; IBM tie line 863-7287 Notes address: Martin W Sachs/Watson/IBM Internet address: mwsachs @ us.ibm.com ************************************************************************************* David Fischer <[email protected]> on 09/20/2001 01:23:08 AM To: Martin W Sachs/Watson/IBM@IBMUS cc: [email protected] Subject: RE: T2 Retry with Delivery Receipt Marty, I really don't want to get into this discussion since it is so far out of scope for ebXML. However, if you are interested, please see the links below. The truth is that no one knows how many transmission errors are undetected and those who have done studies suggest, for sensitive data, exactly what Dan proposed, software checksums. These reports suggest that there is actually a high rate of undetected errors on TCP/IP networks (high meaning there could be the probability of a megabyte file transmission having an even chance of an error). I don't know for sure, nor am I particularly concerned since I can digitally sign anything important and prevent errors through software means. Even a signature is not 100%, but it is close enough. Please, let's not go off on this tangent! Regards, David Fischer Drummond Group. http://citeseer.nj.nec.com/stone00when.html http://www.academ.com/nanog/feb1997/dynamics.html