Hi Jalpesh , you have asked a very good question , from my understanding this is highly subject to the Database system been deployed. But a rule of thumb is , Big data security is increasingly important as security attacks are more common and potentially very costly. This is especially true in the case when there are different types of data which require different security levels. The most commonly used approaches to database security, such as Mandatory Access Control, Discretionary Access Control and Role Based Access Control do not provide truly adequate solutions, as they are either too rigid or not entirely capable of addressing issues relevant to multi-level security situations, such as information inference. Consequently, there is a need for a security solution specifically designed for the situation of databases containing data with different security requirements.

To address the issue of data security in the case of databases containing data with different security requirements, the proposed approach is multilevel, using the following techniques: partition, encryption, integrity lock and sensitivity lock (Pfleeger and Pfleeger, 2006).

Hope this helps and points you in the right direction.

Suggest that you take a look at the the following:

• ARX tool, which can de-identify data for privacy while calculating metrics for the strength of the de-identification as well as the utility of the de-identified data: http://arx.deidentifier.org/
• Some of our experiences and techniques used to protect privacy in big data context:

  Conference Paper Making Big Data, Privacy, and Anonymization Work Together in...

Thanks @philip, @jeff Sedayao, @ Ajit Kumar Roy for answering question.