Forrester Ranks the NoSQL Database Vendors
NoSQL databases have surged in popularity as organizations struggle to serve highly distributed Web applications, but navigating the various types of NoSQL databases and the vendors providing them has proven tricky. Shedding some light on the topic this week with a pair of reports was Forrester Research.
The analyst group looked at two types of NoSQL databases in separate Forrester Wave reports, including document-oriented NoSQL databases and key-value NoSQL databases. Graph databases constitute the third type of NoSQL data store in Forrester’s view, but the group did not chime in on this topic.
Here’s how Forrester breaks down the two NoSQL database types:
- Key-value NoSQL databases are ideal for handling Web scale operations that need to scale across thousands of servers and millions of users with extremely quick and optimized retrieval, Forrester says. The products accomplish this by leaving out many features of relational databases and implementing only features that “extreme Web apps” need around performance, scale, query optimization, availability, and security.
- Document-oriented NoSQL databases sit higher up the complexity scale on the NoSQL ladder, and are optimized for managing semi-structured data in a more efficient manner than their relational cousins. Document-oriented databases are characterized by a flexible data model and are typically used to serve content and documents for next-gen Web, mobile, and cloud applications.
Sounds simple enough, right? Unfortunately, the lines are often blurred between the key-value and document databases. One database may be based on a key-value architecture, but implement some features of a document or a graph NoSQL store. Making matters even more complicated in this fast-moving field is the fact that relational database vendors like IBM, Microsoft, and Oracle are also adopting these types of capabilities too.
But for today the spotlight is firmly on the NoSQL databases and the vendors who develop and support them. The company ranked the leading vendors in both NoSQL categories, evaluating both the product and the company itself.
Document-Oriented NoSQL Databases
In terms of product offering, MarkLogic’s NoSQL database is the best, according to Forrester. “It has the most comprehensive NoSQL document databases data management features and functionality to store, process, and access any kind of structured and multistructured data,” the firm writes in its Wave report. MarkLogic customers most commonly use the database for workloads that mix transactional, analytical, and operational requirements, and is particularly strong with non-structured data, the group says.
MongoDB, on the other hand, wins the popularity contest. “With more than 1,000 paying customers, it’s popular among application developers because it’s easy to rapidly iterate its schema when doing iterative development and deployment,” Forrester writes. Mongo’s application development framework is a hit among developers and its performance and scalability “make document storing, processing, and accessing simpler for large mission-critical deployments,” the firm writes.
Couchbase was confined to the “leaders” category with its hybrid key-value and document-oriented databases, which has support for documents, a flexible data model, indexing, full-text search, and MapReduce for real-time analytics. The product appeals to enterprises “that need both database features and functionality,” Forrester days. However, Couchbase’s offering did not score as well as either Mongo’s or MarkLogic’s, and the company’s strategy fell short to both as well. Couchbase tied MarkLogic for adoption, and questions about its financials clearly affected Forrester’s rating.
Cloudant also landed in the leader’s section with its version of the Apache CouchDB database (separate from Couchbase’s offering). The offering provides hosting, administrative tools, analytics, and support for the core CouchDB, and builds upon the database by adding support for Lucene search and geospatial indexing. The company has deployments across a variety a industries.
Key-Value NoSQL Databases
One of the leaders is MapR Technologies. MapR is best known for its Hadoop offering, but it also offers a database that’s based on HBase, the wide-table key-value store that’s based on Google’s BigTable. MapR’s NoSQL offering ranked highest among the group, with a particularly strong operational component. Forrester seemed particularly impressed with the capability of MapR customers to deploy transactional and analytical workloads in a single cluster.
Despite pressure from startups, Oracle’s NoSQL key-value database, which is based on BerekelyDB, continues to do well, and has more than 1,500 customer deployments, Forrester says. The Oracle key-value store “is a mature, high-performing, scale-out storage engine and provides transactional semantics, fine-grained concurrency, primary and secondary indexes, and high availability features.” Many Oracle customers use the key-value store to complement the relational database in areas of fraud, personalization, and sensor data management, the firm says.
Amazon looms large in the NoSQL world with its DynamoDB, a hosted key-value store that has more than 10,000 customers, according to Forrester. Its reliance on high performance solid state drive (SSD) and its ability to automatically react to performance issues earned it points from Forrester. DynamoDB has been deployed in a variety of areas, including advertising, gaming, ecommerce, powering Facebook apps, and handling sensor and log data.
Joining the leaders is DataStax, which is the commercial vendor behind the open source Cassandra database, which is another wide-row data store based on Google’s BigTable. The Datastax enterprise offering provides additional tools on top of Cassandra, including analytics (it can run MapReduce jobs and the Apache Spark framework), search, monitoring, in-memory, and security capabilities, according to Forrester.
The final leader is Aerospike, which develops the open-source, in-memory key-value store of the same name. “The top use cases for Aerospike include operational, real-time analytics, and transactional that need to run at extreme speed,” Forrester says. The product sports a hybrid in-memory architecture can support DRAM or Flash, and features “a self-healing architecture with zero touch, zero downtime operations,” the analyst says.
Basho Technologies ranked as one of the top leaders in the key-value category. The company behind the Riak open source database scored well with its product offering, which sports multi-data center replication, full-text search, secondary indexes, MapReduce, and a graphical admin. But Forrester gave it a lower score with its market strategy, commitment, and roadmap.
Couchbase rounds out the list of key-value store participants. As mentioned earlier, the company touts a hybrid NoSQL database that combines document-oriented database elements on top of a key-value base. Its product offering scored well but the overall score was dragged down a bit by its go-to-market strategy (which ranked a 3 out of 5, same Basho, Oracle, and Datastax) and its product roadmap.
NoSQL databases are with us to stay. As hyperscale applications and data volumes surpass the capabilities of traditional relational databases, companies have no choice but to alternatives. We’re in the early stages of the NoSQL market expansion at the moment. Forrester estimates the current adoption of NoSQL to be at 20%, but sees it doubling by 2017.