Nosql Databases Live Up to Their Promise?

¶ … NoSQL Databases Live Up TO Their Promise? (Leavitt, 2010) the author presents a fair and balanced analysis of the NoSQL database technologies and supporting Big Data platforms including Cassandra and HBase (the precursor of Hadoop) with accuracy and clarity. The author also points out that despite the cost advantages of NoSQL and its inherent strengths in managing massive amount of unstructured content, its limitations from an enterprise computing perspective including overhead and complexity, consistency including support for atomicity of data and support for the industry-wide ACID framework are lacking (Leavitt, 2010).

The comparison of SQL vs. NoSQL database technologies is well-done in the article, pointing out how NoSQL excels at managing object-, image and textually-based data while the former excels at structured data and the speed of transactions. The article does not mention however how the complexities of managing batch-oriented processes of very large data sets vary between the two database technologies (Stonebraker, 2010). What differentiates NoSQL databases the most however are the key-value stores, approach to defining column-oriented databases and the document-based stores that many of the social networking sites including Facebook, LinkedIn, Twitter and others are predicated (Meijer, Bierman, 2011). These innate strengths of NoSQL databases and the exponential increase in content generated daily on social networking sites will eventually lead to NoSQL being a foundational element of Big Data strategies across the enterprise. Using multiple databases unified by a common data model (Ghosh, 2010).

For all of these strengths in the article however, there are several shortcoming including the following. First, the article doesn't fully address just how much more robust NoSQL is at managing analytics functions and the development, scaling across multiple virtual machines including integration to Sold State Drives (SSDs) as Amazon Dynamo's architecture supports today (Rys, 2011). Second, the role of Latent Semantic Indexing (LSI) and its role in the Google algorithm, parsing of data within the LinkedIn Big Data and data scientist initiatives designed provide greater accuracy of results to searches and ad hoc network analysis are making NoSQL emerge as a more disruptive innovation than SQL's growth (Meijer, Bierman, 2011). Finally in the area of transactions and batch processing, SQL excels in this area due to their inherently more structured approach to data models and interrelationships, while NoSQL is much better suited for analytics operations (Stonebraker, 2010). Even batch-oriented operations of complex queries running on Big Data platforms including Cassandra, HBase and Hadoop will require a measure of SQL-based structure to ensure performance scalability over time (Rys, 2011).

Conclusion