Sergey Nivens - Fotolia
Developers from Internet giants and tech startups have created a slew of NoSQL databases designed to evade the schema rigidity of SQL-based relational software. But their progeny have some issues that can complicate efforts to deploy and manage enterprise NoSQL architecture to support business intelligence (BI) initiatives and big data applications.
NoSQL technologies still have a ways to go in becoming fully enterprise-friendly, at least based on the comments of some attendees at the 2015 NoSQL Now! Conference in San Jose, Calif. SearchDataManagement asked them about the biggest NoSQL problems and challenges that organizations face in working with the databases. The hurdles they cited include difficulties in querying NoSQL databases for analytical purposes and building reports to distribute BI and analytics findings gleaned from the information stored there. Durability and scalability issues also confront NoSQL systems in some settings. The ability to easily create effective data models is seen as another significant missing link for NoSQL platforms.
Karen Lopezdata architect and consultant at InfoAdvisors
In addition, the wide variety of NoSQL databases available in several distinct product categories -- MongoDB, Cassandra, Couchbase Server, Neo4j, MarkLogic, Riak, Redis and many more -- can make it hard to pick one and move ahead with a deployment. For some NoSQL adepts, the answer is using multiple databases to solve multiple problems, which can turn setting up and supporting NoSQL architecture into a multi-headed challenge. In an industry such as healthcare, that can also encompass meeting compliance rules for protecting data privacy -- an issue that may have not crossed the bows of fast-moving Internet companies reaping newfound Web data.
Tassos Sarbanes, data scientist at financial services company Credit Suisse: "Analytics and reporting are the biggest challenges. You can't expect business users to write Java code to extract data from a NoSQL database. In my world, that's a no-no."
Ravi Krishnappa, senior solutions architect at data storage vendor NetApp Inc.: "Over 30 years, we've learned how to write business intelligence applications on top of relational databases -- there are patterns. With NoSQL today, we have no cookie cutters. We don't have any blueprints."
Olga Kirillova, software engineer at movie and TV viewership data provider Rentrak Corp.: "To my mind, the biggest one is to find a system with minimal latency and maximal uptime. The systems sometimes run slowly; sometimes they stop running, even though they're supposed to run 24 hours a day, seven days a week."
Dan Sullivan, independent database consultant and author of NoSQL for Mere Mortals: "I would say it's data modeling -- that is, getting the model down right. There's no point at which there is a sense that 'the model is done and we can move on,' because NoSQL modeling is unlike relational modeling. You look at your queries, and then you design [models] according to that. The problem is that the queries are constantly changing."
Mahesh Chaudhari, software architect at analytics services provider Zephyr Health: "There are different NoSQL databases that can solve different problems. So, we work with more than one -- each presents unique challenges. One of the biggest we see is multi-tenancy. That's because we have to ensure we meet compliance [rules] that protect data."
Karen Lopez, data architect and consultant at InfoAdvisors and blogger @datachick: "I think the biggest challenge is just that there are so many products, architectures and approaches. And because these NoSQL products don't have 30 years of history behind them, it's very hard for enterprise customers to evaluate them."
Despite those and other potential barriers, NoSQL software is finding a place in computing environments in a growing number of organizations. But it will likely have to continue trying to find better enterprise footing in a data management landscape that remains highly focused on SQL and relational databases.
For its part, analyst firm Gartner Inc. sees NoSQL clearly emerging as a widely used technology, but one that will exist within a larger big data ecosystem. By 2017, Gartner predicts, all leading operational database management system vendors will offer multiple data models, both relational and NoSQL, in a single platform.
Discover a book author's tips for building a NoSQL architecture
Find out how users are working with NoSQL database systems
Listen to a podcast on big data analytics and NoSQL challenges