What should a server sound like?

At first you would hear crickets and frogs. Rainfall indicated the flow of bandwidth, and a thunderclap represented a sudden surge in demand. Bird calls would warn of intrusion detection system events, and if a virus had entered the network, administrators would hear a crane.

This collection of “marsh landscape” sounds was the first prototype of a research project at Sheridan College Institute of Technology and Advanced Learning that uses software called iSic (or “information music”) to help enterprise IT systems express themselves in audible ways. After more than two years of effort, it is now in the third week of a six-week trial on the school’s servers, where classical music helps technical staff monitor the servers’ performance. If iSic proves successful – and if the team behind the project can commercialize it as a product – it promises to turn IT management into IT orchestration.

“Sound always has related complex information, whether it’s the braying of a donkey pulling a squeaky cart to scenes to from the starship Enterprise, where Scotty could tell by listening that the warp speed was not right,” said Sheridan Telecommunications Technology Program coordinator Bill Farkis, who along with research fellow Hong Tae Jeon developed iSic. “A network, with all its security issues, is soundless and begging for an opportunity to express itself acoustically.” 

Sheridan uses deep packet data probes located in subnetworks which supply real time metrics to a “super agent” that concentrates data and provides a synchronous data feed to a sound engine. An iSIC server then uses conversion algorithms to process agent data into mathematical music, a Midi synthesizer uses a script to synthesize sound, and music is then streamed through speakers set up near the administrator. The Java-based software, written by Sheridan students, can also graphically depict information about network performance. 

Classical music eventually replaced the marsh landscape sounds, Farkis said, because it didn’t do a good job of indicating the granular changes in a network. “You can’t modulate individual crickets,” he said, adding that the team also tried creating “raw” sounds that could pack a lot of information. “But it was far too synthetic – people wouldn’t tolerate it.” 

Gerard Kaempfe, a system administrator at Sheridan who is one of two staff involved in the pilot, said iSic is made up of piano chords and other easy-on-the-ear instruments. The sound of violins, for example, indicate an increase in spam volumes, which is useful given the college’s complex filtering system.

“It’s really good,” he says of the music itself. “In the back of my mind, I find myself trying to figure out if it’s something I recognize . . . once you become attuned to it, it becomes very easy to identify behaviour.” 

Farkis said he and Tae Jeon are interested in exploring both how much relevant information can be conveyed through “sonification,” and the extent to which the human ear will differentiate it or tune it out. IT equipment has no “native” sounds, he pointed out, which means that in some ways iSic is no more manufactured than the noises we hear on a PC.

“We associate music with pleasure, with dancing and listening and stomping feet. We don’t associate with work environment. That’s really been one of our major obstacles in terms of bringing this to various industries.”

Even if iSic becomes commonplace in data centres, system administrators will probably still need other tools to assist with remote monitoring, Kaempfe admitted.

“I use it as much as possible, but with job description it’s kind of difficult to stay in one place for any length of time,” he said.

Farkis said he and Tae Jeon are seeking an investor to help commercialize iSic, as well as another test environment to pursue their research further.  

Comment: info@itbusiness.ca

Share on LinkedIn Share with Google+