Network-Monitoring Data Put to Music

StrongGlad writes "Building on the idea that people are naturally attuned to sound, the Sheridan College Institute of Technology and Advanced Learning has created software that translates network and server activity into music. And, their IT department operators can interpret the music to detect problems in the system." Talk about finding the beauty in Spam. From the article: "Last Friday, IT department operators began listening to what sounds like classical music but is actually a precise audio model of system metrics. They are trained to recognize instruments, chords, tempo and other musical elements of music as a translation of e-mail activity from 15 servers over three subnets. Every aspect of the music correlates to information. Probes detect server activity and send about 20 summaries a second to the iSIC sound engine. The data is aggregated and transformed into an audio format."

Been done before?

[Narcissus]

I swear I remember reading about something like this years ago but for the life of me, I haven't been able to find it mentioned anywhere.

Although it wasn't email / spam related, the system I'm thinking of used jungle sounds (birds, rivers etc.) but had things like lion roars when the firewall detected a hack attempt.

Am I just dreaming this, or can someone give me any more information?

Something like this

[db32]

I have seen something about a similar project that used graphical patterns and colors/intensity/patterns indicated potential problems. I think this would probably be alot nicer since it doesn't leave you staring at a monitor all day (yes I know most of us do this anyways). With networks getting larger and more complex things happening on them, projects like this are definetly an interesting avenue for monitoring. I know people that can read tcpdump screens at a truely disturbing rate, but being able to sit and "watch" all the logs of everything in their multitude of formats and indicators is going to be a huge leap forward in detection and prevention. Most intrusions aren't caught until well after the fact, if at all. Having something like this that could potentially alert admin and security folks of trouble on the network, malicious or not, would be awesome.

Been done before...

[Linker3000]

Safety systems in some installations handling radioactive materials broadcast a background sequence of notes/clicks (*not* anything like a geiger counter) through loudspeakers in critical areas - the 'melody' is designed to be unobtrusive under normal conditions (your mind 'tunes it out'), but the notes change under alarm conditions or when certain monitored values start moving and even minute variations in the sound are immediately obvious to those in earshot. This has been in use for tens of years. ..and some of us just have to stare at a Nagios Web page or wait for an email that triggers a 'beep' sound.

Re:Maybe we can finally answer the age old questio

[Janitha]

Well to listen to slashdot, you could send it into a audio device. As root wget -nv http://www.slashdot.org/ [slashdot.org] && cat index.html > /dev/audio Or you could ghetto rig the machine to output the network dump such as from tcpdump directly to a audio device, not as nice music as the original post, but it will work.

This idea is absolutely bunk!

[Anonymous Coward]

Incredibly frelling dumb. More in one way like Smithers than SMITHERS himself. What kind of person calls a SYN packet meaning F# and an echo a middle C. Just because you can make notes doesn't mean that it's music...
 
If you get a zebra to jump on a keyboard and play notes, how is that music?
 
Much training can cause interpretation of anything. There are people who can understand speech from a frelling frequency diagram. The nerve.

silence vs multitasking

[DeveloperAdvantage]

Demarco and Lister's Peopleware book has a good section on the importance of a quiet workspace. In a study they quote (this one from Cornell in the 1960s), researchers split a group of computer science students into two groups, the first group listened to music through headphones and the second group was in a silent room. Each group was given the same programming problem, which consisted of a series of mathematical operations, to implement from a specification. The speed and accuracy of the programming was about the same in each group, but, the assignment itself was a trick question - the end result was that the output number was the same as in the input. And, of those that realized this, the overwhelming majority came from the quiet room.

Most "technical" work uses the left side of the brain, I suppose leaving the right side of the brain free to listen to music to monitor the system. But, every so often, even in what is considered "technical" work, a person needs to be creative, and it would be unfortunate if at that point in time your right side of the brain is off monitoring the system.

Of course, if multitasking is so important, audio content is really the only content which has the potential for effective multitasking.

Been doing this for years..

[stevied]

Anyone who's been sat next to a noisy server has probably been doing this for years. I found I became rapidly attuned to normal disk activity patterns, and could detect unusual goings on very quickly.

I also used to be able to recognise the connect speed of analogue modems by listening the negotiation, but that was many moons ago..

You guys might be missing the point

[Knertified]

I think the point of converting logs to music is the human ear is better at picking up patterns than a pair of eyes. If you hear a melody in the beginning of the day, and hear that same melody at the end of the day you will recognize it. A melody could be 200 lines of sequential log. I doubt someone could visually remember 200 lines of log.

I think this is an excellent idea.