Most companies don’t get rid of the currently unused data, with the expectation it may be required to forecast trends in the future. This data becomes ‘dark’ data when the company completely forgets its existence and it becomes difficult to locate it and interpret it or interact with it. However, this ‘dark’ data can be restored or ‘lightened’ up by the following steps:
1. Data filtration, that can isolate the required information in the means of filtering the data feeds, to prevent the accumulation of excess unnecessary data, but is only useful when the data center manager can identify the data, expected to be valued in the present or future times.
2. In case of any future requirement, it is always better to export the data (instead of deleting it) to a cloud-based storage system, from where it is possible to import the data into the company’s data system, whenever required.
3. It is important to properly define, data withholding policies in order to facilitate data center management.

Read more at:http://www.techrepublic.com/article/clear-out-dark-data-to-make-room-for-useful-big-data/