Time to start looking at treatment of data waste

Computer scientists say it's time to start looking at treatment of data waste

As anyone who has ever used a Windows based computer for any length of time knows, the longer you have it, the slower it goes; this is because of the accumulation of data files and entries in system logs; information that in many cases isn't really necessary. Thus, our computers slow down due to the accumulation of "waste." Now, two computer scientists from Johns Hopkins University have published a paper on arXiv, where they argue that data waste management on computer systems could, and should be handled similarly to the way physical-world waste is managed.

In their paper, Ragib Hasan and Randal Burns pick up where computer scientists at Cornel University left off after discovering in 1999 that up to 80% of files written to the hard drive by the Windows NT operating system were deleted within five seconds of being created.

Hasan and Burns analyzed three computers: a MacBook laptop, a desktop running Ubuntu Linux and a Fedora Linux fileserver in the University Library (Linux is a variant of the Unix operating system used primarily at educational and research institutions). Their intent was to find out what percentage of the files on each of the computers had not been accessed since their creation. They found that the percentages for each were: MacBook: 20.6, Desktop: 47.4 and Server: 57.1 and that the percentage of disk space used for each was 98.5, 38.1 and 99.5 respectively; clearly indicating that a large number of files using a lot of disk space had never been used again once being created. This is clearly an inefficient use of resources.

It is for this reason that the duo suggest a new approach be used for data waste, one that takes advantage of the research already done with physical waste; specifically, they suggest a pyramid approach be used, similar to the one put in place by physical waste management companies. At the bottom of the new pyramid would be the worst case scenarios, then moving up, the next best and so on till reaching the top, and that they be labeled as such: Dispose, Recover, Recycle, Reuse and Reduce, with zero data waste being the optimal goal.

Read more.

The paper: The Life and Death of Unwanted Bits: Towards Proactive Waste Data Management in Digital Ecosystems