According to Forrester research, NoSQL databases continue to play a prominent role in the enterprise IT infrastructure. Forrester estimates the current adoption of NoSQL to be at 20% and is likely to double by 2017. Forrester defines NoSQL as:
A non-relational database management system that provides storage, processing, and accessing of complex data structures and support for large volumes of polystructured data. It supports a horizontal, on-demand, extreme scale-out database platform that delivers a schema-less and flexible data model, and is optimized for high performance.
The following are, according to Forrester, applications where enterprise architects consider using NoSQL databases:
- Operational databases supporting new generation operational analytics, including real-time and predictive analytics.
- Stream processing that needs to scale across many nodes in a clustered configuration.
- Databases supporting low latency ad hoc queries for mission-critical applications.
- Management and storage of increasing volumes of structured and unstructured data.
Traditionally NoSQL databases have been categorized into three main groups - key-value stores, document and graph database. However, Forrester finds that the lines between these categories are being blurred by many vendors as they seek to create database platforms that can satisfy the broader needs of enterprises and widen their appeal to application developers. For example, AWS DynamoDB just announced support for storing JSON documents.
Forrester’s evaluation is on 57 criteria, which are grouped into three high-level categories:
- Current offering. The breadth and depth of each vendor’s NoSQL key-value database product set, based on each solution’s architectural and operational functionality.
- Strategy. The way the vendors plans to evolve their NoSQL key-value database solution to meet emerging customer demands including their go-to-market approach, commitment, and direction strategies.
- Market presence. This includes vendor’s company financials, adoption, and partnerships.
According to Forrester, the following are the current leaders in the NoSQL space - Aerospike, Amazon Web Services, DataStax, MapR Technologies, and Oracle.
Forrester also defines Couchbase and Riak (Basho Technologies) to be close next by offering competitive functionality.
Based on Forrester’s evaluation, Amazon’s Dynamo is leading in strategy and market presence while MapR is leading the pack in current offerings.
According to Forrester, Amazon DynamoDB leads as the top NoSQL key-value database in cloud offering options:
Amazon DynamoDB is a fully managed NoSQL key-value cloud database platform that uses solid state drive (SSD) to store, process and access data delivering high performance and scale. It also offers automatic sharding of data to scale across servers to support larger high-performance application demands. A differentiating performance feature is that Amazon Web Services actively monitors the performance levels of the NoSQL database service and reacts to any sign that performance has degraded. Enterprises use Amazon DynamoDB to support advertising campaigns, drive Facebook applications, track gaming information, collect and analyze sensor and log data, and scale eCommerce applications. Amazon Web Services has tens of thousands of customers using Amazon DynamoDB, some with very large implementation.
MapR Technologies offers broader NoSQL databases capabilities with the highest throughput:
MapR provides a native key-value storage engine based on Google Bigtable. The data model is known as “wide-column” and includes columns, timestamps, and column families. MapR provides both declarative and imperative languages for data query access. MapR-DB is an in-Hadoop database that integrates natively with Hadoop. It supports automatic sharding and re-balancing of the cluster to support broader scale. MapR customers have deployed all types of workloads including transactional, analytical, predictive analytics, and mixed. Since MapR provides NoSQL key-value integration with Hadoop, MapR customers often deploy a mixed workload in a single cluster. MapR Technologies road map focuses the key-value store to be further integrated with the Hadoop ecosystem. In addition, it plans to further extend its features on ease-of-use, improve on performance and scale, and support a larger distributed in-memory data platform.
Forrester’s report helps people interested in using NoSQL databases to evaluate their options and to make more informant opinion on what is available.