Re: Question about hashing algorithms
Available news archives: comp.lang.tcl - comp.lang.python - comp.security.firewalls - sci.crypt - comp.lang.php - comp.lang.javascript
Google
 
Web news.hping.org


sci.crypt archive

Re: Question about hashing algorithms

From: Unruh <unruh-spam@physics.ubc.ca>
Date: Fri Aug 26 2005 - 23:42:17 CEST

daw@taverner.cs.berkeley.edu (David Wagner) writes:

>>Are any hashing algorithms that are secure for short data and have
>>short signatures? As I understand it, SHA-1, MD5, etc. were pretty
>>much designed to work for any size of data. Sometimes that seems like
>>overkill, though.

>What's wrong with overkill? (a serious question)

>>If I have small bits of data (say, a few kilobytes
>>or so) and I want to verify each chunk with a hash, it seems a little
>>wasteful to require a 160 bit hash that matches up with what might only
>>be 2K of data most of the time.

160 bits = 20 bytes. What is wrong with a 100-1 ratio?

And security relies on length. If you ahve a one bit hash it is trivial to
find collisions and preimages attacks. the longer the hash, the harder,
other things being equal.
Received on Thu Sep 29 21:51:00 2005