Data security

Spam vs. ham ratio

I just did a little survey on my own. My spamfilter is set up in a way that it logs each letter into a logfile (spam and ham both into their own logfiles) and keeps record of some of the emails' properties (From, To, size, matching Spamassassin filters, date, ...). The spamfilter works pretty well: 95% of real spam is identified as spam and only a very small number of emails (let's say 0.1%) is a false positive (where I mean a ham is identified as spam). According to my logs it seems that most of the time only 1/3 of all incoming mails is ham and 2/3 is spam. Around Christmas and New Year's Eve things got a lot worse: 4/5 of all my incoming mail was spam! Shocked Thanks to Spamassassin I could still enjoy reading my mail even during Christmas. Smiling

How to create (or verify) an SFV in Mac OS X

An SFV file is just a simple text file with each line containing a filename and a corresponding CRC-32 checksum. On Windows you can easily generate this file with a number of GUI apps (eg. Total Commander), but on the Mac you're left in the cold. There's one app named MacSFV, but to be honest, I could not find the DMG for the latest 2.0.1 version (dated back in 2003) on the net. Shocked

Minimal Key Lengths for Symmetric Ciphers to Provide Adequate Commercial Security

I just read the PDF on the subject. It was written a while back, but most of the conclusions stand still today. However my mantra is to use always the highest available key length that your tool supports since you can never be cautious enough. Just remember how secure people felt using DES for a long time. Looking at the computing power that is at hands today, it's merely a joke to crack a DES encryption. I assume that some people on Earth are always at least ten years before the masses, so if you're looking for security today, then think ten years forward. Secure your data basing on history of computer security. Assume that scientists will find ways to scale down brute-force attacks against currently available alogithms to a 1/100th (or even 1/1000000) of their known complexity. Protocols and algorithms get compromised every day, do not make the mistake of trusting a single piece of code/logic/alg. As an old Hungarian ad said: "one lock does not lock". Smiling

Syndicate content