Industry executives and experts share their predictions for 2019.  Read them in this 11th annual VMblog.com series exclusive.

Contributed by Krishna Subramanian, COO, Komprise

How AI will Shape Data Management in 2019

Artificial Intelligence (AI) is all the buzz these days, and vendors of all types are touting Ai in their products. But with no standards or clear definitions of what constitutes AI, customers are right to be skeptical that AI is just the latest marketing buzzword.  There are three clear and tangible ways in which AI can simplify data management and watch for vendors to evolve their uses of these AI principles in 2019.

1.       Adaptive Automation via Machine Learning

No two customers' data sets are exactly alike, nor are their infrastructure or user needs. So why should software operate in the same way in all environments?  By applying analytics and machine learning, data management software can observe the unique constraints of your environment and "learn" to work smarter by adapting its algorithms to your needs.  For example,  today's modern calendars don't just remind you fifteen minutes before any appointment, they instead let you know based on the appointment location and traffic when you should leave. Similarly, data management software can move and analyze data based on your network, storage and user patterns.  Based on your network and storage load and the sizes of your data, it can figure out the best order and times of when to move data. It can tell you based on your growth rates when you are likely to run out of space well before it happens. Watch for this type of adaptive automation to become more advanced in 2019.

2.       Goal-based versus Rules-based policy management

Traditionally, data management policies have been rule based - an IT administrator programs a set of rules on how the software should function. The problem with this approach is that a human has to predict every single eventuality and program a rule for it.  Rules could conflict with other rules, which create problems.  This approach is very tedious, error-prone and time consuming.  But, imagine if an IT administrator could set goals e.g. I don't want anything that hasn't been used in over a year on expensive storage, or I want these users to get the highest performance for their data, etc.  And based on your goals, the software figures out the appropriate way to achieve them.  This type of goal-based policy management ensures IT is in control of the outcome they desire without the manual heavy-lifting and errors of rule-based approaches.  Watch for more sophisticated goal-based policy management that combines analytics, tagging, and goal-based execution.

3.       Big Data Analytics to Extract Data Value

A key roadblock to adoption of big data has been the difficulty in finding the right data sets to analyze, since data is strewn across billions of files and different storage silos, both on premises and cloud. For instance, let's say you want to find all files related to patients who had a particular type of cancer, imagine if you could get a virtual data lake on demand of all this data no matter where it lived so you can then analyze it.  AI can drive efficient search and discovery of data to help us extract value from our data more efficiently, even at today's massive scale of data.

Finally, one trend we believe you should never subscribe to in your data management solutions is Autonomous or automatic management.  An autonomous or self-driving system automatically makes decisions on your behalf - and while this might be good for a self-driving car whose goal is to get you safely from point A to point B, data management is far more nuanced. There isn't one correct goal for all scenarios and it is important that IT maintains control and can guide the outcomes - so beware of products that claim "fully automated management" on your behalf! 

##

About the Author