Big data analytics is among the current top three concerns of IT leaders, other important areas being IT Security and Mobile Device Management. With increasing thrust of organizations for catering more information, the size of data to be handled is getting bulkier. The idea of ‘Big data’is not mere expansion of traditional database but a lot more. It is not just data maintenance but also analysis and maneuvers of huge unstructured data using predictive techniques and other mathematical models and its conversion into insightful information adding business value.

Data being the most important and vulnerable asset for organizations is critical to kept secure. All the organizations are growing in terms of data they handle. From the inventory stocks to target and existing customer base, from social media content to web locks, each and every bit is adding to data to be stored and analyzed. Thus, leaving them with an only option of switching to Big Data. But, Big Data breaches will be big too, with the chances for even more serious reputational and legal damage than at present.

Big data is often characterized by 3Vs, Volume, Velocity and Variety of data. All the three attributes have their own challenges associated with them. Massive size of data is off course most prominent and coherent one. Velocity signifies both, rate at which data arrives and needs to be acted upon. Variety denotes the heterogeneous data, unstructured data and the issues related to it.The concept of Data has been highly evolved than the traditional times and so are the techniques to handle them. There has been a paradigm shift from traditional Relational databases where data was put into processor to BIG DATA where multiple processors are brought to the data. A technological shift has supported this change to happen. Platforms like Hadoop and frameworks like MapReduce and Storm. Several others technologies for real time analytics and graph computation are emerging as Hadoop has already proved unsuitable for many existing problems.
Big data has resolved several issues like processing of adhoc queries and enormously flooding data through its parallel and powerful computational frameworks .Its distribution across variant silos and ability to tackle unstructured data makes it way superior than traditional analytic techniques. These scalabilities in turn lead to several bottlenecks for the analysts to handle. Following are certain vulnerable areas of BIG DATA which might lead to challenging situations if not tackled well.

Heterogeneous Sources
End Point Input Validation is important to ensure integrity of data sources. Due to massive amount of data to be dealt, it becomes more challenging and increases the vulnerability.

Insecure Data Storage
Data being stored at thousands of nodes, secure storage ensuring Confidentiality, Integrity and Availability of stored data itself becomes challenge.

Insecure Data Processing and Computation
Untrusted computation programs can be submitted and used by the attacker to extract critical information from data sources or to manipulate the sensitive data. Several attacks like denial of service can also be initiated by the attacker.

Data Mining and Analytics leading to Privacy Breeches
Monetizing of big data needs data mining, analyzing and ultimately sharing of results. This might lead to privacy breeches, invasive marketing and disclosure of sensitive information.

Cloud Adoption
Switching to Cloud for storing enormous data proves to be a boon but at the same time it outsources the security of the owners’ critical data to a third party and hence becomes vulnerable.

Insecure Access Controls
Big data implementations include open source code leading to unrecognized back doors, intruders and default credentials. Also, authentication and access from various sources may not be secure enough.

Auto- tiering
It creates two repositories of data ‘Hot data’ and ‘Cold data’. The data which is accessed less frequently is moved to Cold Data which is lesser secure medium. This might be risky if the cold data is sensitive one.

Compliance Monitoring
Real time monitoring of compliance is a big time issue that should be taken care of due to the excessive amount of data and alerts generated by big data.

Audits
Regular audits and governance become even more necessary yet challenging due to highly distributed and unstructured nature. Timely audits in conformation with universal compliances are indispensible to ensure that the data is not yet compromised.

Big data was initially thought and designed with a concept of scalability i.e. accommodating huge volume and variety of data and responding to the queries at earliest. Security issue was not much thought of, which is lacking and need to be made scalable with the amount of data being accommodated.

Sonika Singhal
MBA-IT
IIIT Allahabad