Adoption rates for NoSQL database technology and its deployment are relatively modest at present; Forrester recently indicated this technology occupies 20 percent of the overall market share. However, there are a number of trends in Data Management that will further spur those adoption rates in 2015, including:
- A blurring of the distinction between NoSQL databases and SQL databases
- A blurring of the distinction between NoSQL database functionality
- Increased support for JSON
- The Internet of Things
Amalgamation of NoSQL Database Functionality
Originally, NoSQL databases were characterized into four different types: document stores, key-value stores, graph databases, and wide column databases. More frequently, databases belonging to one of these four traditional types are taking on the characteristics of others within the NoSQL categorization. Gartner commented on the emerging trend of multi-model NoSQL offerings:
“Multi-model DBMSs combine the characteristics of multiple NoSQL models in a single deployment footprint. Instead of selecting a single data model per platform, such as table style or key value, DBMSs with several data models may offer table-style, key-value document stores and graph DBMSs on one platform. This approach allows multi-model DBMSs to address licensing, architecture, deployment and implementation in a single footprint — making multi-model DBMSs ideal for single applications requiring multiple data model.” (Heudecker and Adrian, 2014)
Each of the individual NoSQL database types offers the same basic advantages—extreme scalability based on a horizontal distribution of data and schema-free models to accommodate a multitude of data types and structures. Due to the ways that they actually store data, however, key-value databases are preferable for Web scale operations while document stores are advantageous for management of semi-structured data.
Nonetheless, the tendency towards amalgamation is evinced in Couchbase, a hybrid document-store and key-value store, Oracle’s NoSQL platform, and FoundationDB. The fairly recent announcement that Cloud-service provider Amazon Web Services’ key-value database DynamoDB supports JSON document storage further reveals this trend. 2015 will see more of such announcements as well as a refinement in the capabilities on the part of vendors across NoSQL database types.
SQL’s influence on NoSQL will be both profound and beneficent in the coming year. The trend that began in earnest when SQL vendors began supporting platforms such as MapR’s Hadoop to take advantage of its Big Data capabilities will only increase as SQL vendors continue to issue support for SQL databases for a variety of purposes. Highlights from a recent Market Research Media report focused on NoSQL state:
“Relational Database Management Systems (RDBMS) are here to stay. Non-relational databases are not a replacement but rather a supplement to RDBMS. We forecast gradual convergence of both technologies into a hybrid ecosystem and takeover of NoSQL technology leaders by established RDBMS vendors.”
The merging of the capabilities of different NoSQL database types will be mirrored (or perhaps is actually mirroring) the merging of NoSQL and SQL technology in relational databases. Gartner has even gone so far as to predict that, “By 2017, the “NoSQL” label will cease to distinguish DBMSs, which will reduce its value and result in it falling out of use. By 2017, all leading operational DBMSs will offer multiple data models, relational and NoSQL, in a single platform” (Heudecker and Adrian, 2014). Part of the interest in combining these two technologies pertains to renewed interest in key-value stores and their utility for large sets of data—particularly those pertaining to fraud, sensor data management, and other applications.
Regardless, the enduring relevance of SQL in regards to the number of legacy systems, overall comfort and training of those who utilize such databases have made it all but inevitable that NoSQL join it, and not displace it, in continuing to provide value to the enterprise. Look for conventional SQL vendors to offer more integration and support for NoSQL technologies. Eventually, such support will exist across all the types of NoSQL databases. Such seamless integration, however, will likely take place beyond 2015.
Although the trend to incorporate NoSQL functionality into relational databases is growing, there is a definite focus on certain technologies more than others. Interest in the capabilities of document stores is the most prevalent among relational vendors that have added NoSQL functionality, with support for JSON increasing in popularity. Although other NoSQL databases have provided support for JSON, document stores are most noted for this functionality (in addition to supporting XML) for Web-based applications. A number of major vendors (Oracle, IBM, and Amazon) have introduced JSON support; such support will soon become a routine characteristic of relational databases by the end of 2015 or 2016 at the latest.
The popularity of JSON among relational vendors reflects the overall presence that document stores have throughout the NoSQL community, which predominantly includes developers. The advantage of using document stores is that they are ideal for expediting the process of application development—particularly apps that apply to the Internet, mobile devices, and the Cloud. The flexibility of schemas that this technology provides enables developers to quickly add features and functions to applications, much more quickly than they could with relational options. Combined with the capacity for NoSQL options to determine the rate of replication of data changes between nodes throughout a cluster, these aforementioned aspects of document stores contribute significantly to their utility for developers creating applications for the most cutting edge Data Management technologies. Gartner revealed:
“Based on Gartner inquiries, partnerships and total venture capital investment, as well as overall momentum… the market continues to be driven by document-style DBMSs, particularly MongoDB. Its developer-centric approach has made it a popular first NoSQL DBMS.” (Heudecker and Adrian, 2014)
Internet of Things
NoSQL applications are widely responsible for the processing of Big Data in real time for the purpose of analytics and operations of mission critical business objectives. The Internet of Things (IoT) is the acme of Big Data technology in that it requires connectivity between devices to derive both action and insight. The projected influence of the IoT will make strides in the coming year, which will help shift the focus on non-relational technology that can accommodate it. The geographic distribution of data and computing resources that NoSQL databases (especially when combined with relational options) provisions is essential to the IoT, and will help solidify the presence of NoSQL technologies across vertical industries.
The future of NoSQL technologies cannot be secure with the increasing adoption rates in Big Data and its numerous applications alone. It also partly hinges on the ability of vendors and the enterprise to address some of this technology’s weaknesses, which have traditionally included:
- Transactions/transactional data
- Data Governance
Combining NoSQL technologies in relational offerings can address some of these drawbacks, particularly their overall effect with transactions, governance, and security issues. Yet, vendors must take dedicated efforts to address these pain points for this technology to fully emerge from the back offices of developers and to become a mainstream technology.
2015 trends in NoSQL will be based on integration. This integration will take place among the different varieties of NoSQL technologies, and between SQL and NoSQL options. The ease and effectiveness of that integration will determine the rate at which NoSQL will be adopted throughout the enterprise, and not merely by developers creating applications. Although this trend will solidify in 2015, mainstream adoption of (and integration of) NoSQL will not likely take place until the end of the decade. According to Gartner:
“DBMS architectures supporting multiple data models, or multimodel DBMSs, have not yet captured mind share. However, additional features from DBMS leaders and efforts by NoSQL innovators to move their products into adjacent markets will raise the profile of this new model in 2015, potentially simplifying the use and operation of these data stores.” (Heudecker and Adrian, 2014)
The coming year will involve more announcements from vendors regarding their incorporation of NoSQL technologies, as well as more announcements of NoSQL vendors procuring significant amounts of capital.