Cognitive Storage Empowers IT Systems With Data-Value Analytics
Data scientists at IBM’s Zurich labs have announced a research initiative meant to transform databases into cognitive storage centers — advanced information systems that can evaluate and make decisions about how to keep and protect the data they contain.
A New Era of Machine Learning
According to ComputerWeekly.com, the plans represent an evolution in how machine learning will work with IT infrastructure. Cognitive storage systems will assign value to different types of data, select the best media on which to save it and be charged with decision-making on governance, data protection and data lifecycles.
As the new technology is developed, one of the key benefits will be lowered costs of storing data. As The Stack notes, secure, highly responsive data storage that operates with minimal latency can be one of the most expensive parts of an IT framework.
IBM’s new cognitive model would streamline and redefine the way hardware, staffing and energy resources are used. The most frequently accessed data in a set would be assigned the highest value and then be stored on the most responsive media, kept in multiple backups and given the greatest security attention. Meanwhile, low-demand data would be assigned to less expensive media for long-term preservation. All of this would occur in the context of regional data governance rules, so regulated information would not be accidentally de-prioritized, even if it is accessed less frequently.
Cognitive Storage Bodes Well for Enterprises
The development of the new system emerges in part from IBM’s work on the DOME initiative, a radio telescope project currently under design by Astron. According to The Stack, the project is expected to ingest up to a petabyte of data daily, and the value assessment and sorting functions of the cognitive storage algorithms stand to augment this process in powerful ways.
The prototype system also promises advantages for enterprises. As IBM researcher Vinodh Venkatesan told ComputerWeekly.com, “For an enterprise, there are ‘must-keep’ classes of data, and these could be set to be of permanently high value. … The rest, the majority, which cannot necssarily be manually set, can be handled by cognitive storage — such as big data-type information and sensor information that might have value if analyzed.”