On Thu, 2002-12-12 at 09:04, Richard Welty wrote: > > However, I wonder if MD5 will acutally help in court as long it is not > > protected inside a crypt sig - somebody out there with an opinion on > > this? MD5 is sufficient to demonstrate the integrity of the message. How would you actually go about proving the authenticity of the hash? What constitutes authenticity of a hash produced from a log file? The purpose of the hash is to show the log file hasn't been altered since the hash was produced. It doesn't matter who calculated the hash, as long as you can make the court believe it was done *before* anyone got a chance to tamper with your log. The only party that could vouch for that is the log source itself. It would make sense for the log source (the syslog* client) to sign each and every log record before sending it out, the signature being an MD5 or SHA-1 or similar hash of the record, encrypted with the source's private key. The price of this is some substantial PKI infrastructure overhead. Alternatively, if you've got a trusted data channel between log client and log server, the server could do the signing of records. Either way, you'd get a log file that carries a signature for every line. The signature can be transformed into the original hash by simple decryption. > depends on the circumstances. in normal deployment in IPSec or SSL, you > would use HMAC-MD5 or HMAC-SHA1, which are signed, and thus provide Actually, they're not signed but keyed. They are only meaningful to parties who share a secret key. > authentication of the source as well as modification detection. It only allows those functions to be performed by someone who also possesses the secret key used in HMAC-*. That means in court you'd have to disclose your secret key. Usually, keyed hashes are transient, use session keys of limited lifetime and get thrown away after verification. (A notable exception to this is DNS TSIG.) Cheers Steffen. _______________________________________________ LogAnalysis mailing list LogAnalysisat_private http://lists.shmoo.com/mailman/listinfo/loganalysis
This archive was generated by hypermail 2b30 : Wed Dec 11 2002 - 17:14:58 PST