Welcome!

Recurring Revenue Authors: Yeshim Deniz, Zakia Bouachraoui, Elizabeth White, Pat Romanski, Xenia von Wedel

Related Topics: @CloudExpo, Containers Expo Blog, @DXWorldExpo

@CloudExpo: Article

Findings on Database Management | @CloudExpo #Cloud #IoT #BigData

Technical decisions around data persistence are hard, which is why we surveyed 583 IT professionals

Technical decisions around data persistence are hard, which is why we surveyed 583 IT professionals on everything from current DBMS and ORM usage to modern database engines' data structures and access patterns to storing data on a mobile device.

The demographics of this survey are as follows:

  • 69% of these respondents use Java as their primary programming language at work.
  • 68% develop primarily web applications.
  • 66% have been IT professionals for over 10 years.
  • 45% work at companies whose headquarters are located in Europe, 27% in the USA.
  • 44% work at companies with more than 500 employees, 19% at companies with more than 10,000 employees.

Give the key findings below a read and let us know what you think.

Oracle, MySQL, and SQL Server Remain Head And Shoulders Above the Rest; Oracle and MySQL Remain Neck-and-Neck
The two most mature commercial DBMS offerings (Oracle and MySQL) are used in production by 51% and 49% of respondents, respectively-significantly ahead of the third-ranked DBMS (SQL Server, at 34%). The top three, and the tight race between the top two, have not changed in years, among our survey respondents as well as on the DBMS ranking aggregator dbengines.com. The nearest NoSQL challenger, MongoDB, remains a distant fourth in production environments.

NoSQL - Especially Document-Oriented - DBMS Adoption Is Significantly Greater in Nonproduction Environments
Non-production environments are more friendly to less mature and less thoroughly supported database management systems and also more likely to be affected by desire to optimize for structural fit and ease of access. In production, where data stores are often managed by specialist non-developers, factors other than developer experience and optimal match between data processing and storage and retrieval algorithms weigh into DBMS selection more heavily. NoSQL and generally less mature and/or less supported offerings should therefore be more popular in non-production environments. DBMSes that implement simpler storage models well-suited to lightweight prototyping - especially, therefore, document-oriented DBMSes - should gain an extra boost in development environments.

Accordingly, the gap between production and non-production usage is greatest for the two most mature commercial DBMS offerings: Oracle (at 51% in production vs. 37% in nonproduction) and SQL Server (34% in production vs. 25% in non-production), and the gap between the most popular NoSQL offering in production (MongoDB, at 20% adoption) and the least popular of the top three (SQL Server, at 34%) enters within the survey's margin of error in non-production environments (where MongoDB enjoys 25.4% adoption vs. SQL Server's 24.6%).

MongoDB's (static-schema-free) document orientation, familiar JSON-like document format (ordered lists supporting a variety of types), and widespread connector availability make it easy to set up without heavyweight data modeling and relatively straightforward to use for many less-data-intensive applications without cramping application architecture or code. Indeed, many non-relational stores are easier to spin up quickly than a full-power RDBMS. Some benefits of the relational model (especially integrity enforcement) are less relevant in nonproduction environments, where updates don't always need to propagate across all entities.

Note: of the top three DBMSes, only MySQL enjoys greater adoption in non-production vs. production environments. MySQL is especially likely to be many developers' default nonproduction RDBMS, presumably because it is popular, open source, mature, familiar, and supported by a strong community. (For the importance of familiarity in developers' preference for a particular data persistence technology, see the upcoming section on matching storage model to data structure.)

Applications Are Almost as Likely to Use Two Storage Models as One
Developers may use more than one storage model in different applications with no reference to the work done by the application; variety by developer speaks more about the human than about the technology. But variety of storage models within a single application indicates "polyglot" persistence - that is, how many storage models are used to persist data where technical and business needs overlap. Among our respondents, nearly as many respondents typically use two storage models in their applications (38%) as use one (40%). This result confirms that "NoSQL" is better understood as "Not Only SQL" because the most popular storage model (given DBMS and query language usage data) remains relational. Based on DBMS adoption data, the second most popular storage model by user count is probably document-oriented; but because other storage models (especially column-oriented, graph, and key-value) are particularly well suited to analytical processing of many data rows, further research is required to discover storage model usage by data volume. In any case, the near-parity between one and two storage models per application indicates increasing interest in matching persistence mechanism to the structures of data to be persisted.

Matching Storage Model to Data Structure: Modeling Graph Data
Graph structures do not fit the relational model comfortably. In a relational database, most (Shannon) information is stored in the columns and rows of each table; the schema is a technical construct designed to enforce data integrity, make the data model more legible, and make the querying model more efficient; not to encode more information. In a graph, however, most information is stored in the structure of the nodes and the edges; additional information about nodes and edges is treated as metadata. Yet many real-world entities are most naturally represented as graphs: social, travel, and trade networks; packet routes; control flows; etc. Storing graph structures in tabular storage is inelegant and inefficient even at first, static only glance; but the problem gets worse in a dynamic setting. Because a graph's computational complexity may diverge wildly from its combinatorial complexity, reducing a graph to a relational schema (e.g.,  two-column mapping tables that relate a row in one table to a row in another - that is, modeling nodes as columns and edges as rows in a new table) may work far better for some algorithms than for others (in ways that are not immediately obvious from the graph itself).

Nevertheless, three factors encourage developers and DBAs to store data that is naturally modeled as a graph in a relational DBMS: first, the maturity of relational DBMSes; second, the simplicity and familiarity of SQL (which 90% of respondents use regularly); and third, the availability and maturity of powerful object-relational mappers (ORMs) that make relational data easily accessible (often with automatic and highly effective optimizations) from application code.

Accordingly, only a small minority (20%) of respondents persist data that is naturally modeled as a graph in a specialized graph DBMS. Further, more respondents store naturally-graph data in a relational database without explicit modeling of edges as rows (39%) than with node-node mapping tables (31%). We expect this distribution to change as graph DBMSes and query languages grow more familiar, as tooling ecosystem around these DBMSes approaches the maturity of ORMs, as inefficiencies introduced by storage-structure mismatch grow more expensive as graph data volume increases, and as use cases (and corresponding storage and retrieval algorithms) grow more varied.

Two possibly linked correlations are also worth noting. First, the largest chunk of respondents who store graphs in a relational database without explicit modeling of edges use Oracle (25%)-probably the most mature and most thoroughly optimized RDBMS. Second, the largest chunk of respondents who store graphs in relational database WITH node-node mapping tables use MySQL (24%), which is also the only RDBMS that gains popularity in non-production vs. production environments. This difference may be a function of both the greater likelihood that MySQL will be used for experimental purposes - where graph problems, insofar as conceptually farther from actuarial use (for which relational databases are a more natural fit), are more likely to appear.

Matching Processing Approach to Storage Model: Use and Enjoyment of Orms
Most developers use SQL (90%) but the relational algebra does not naturally capture object-orientation. Objects do not fall into Venn diagrams; but objects and relational tables do share enough structure that, for many simple (few-join) access patterns, the so-called object-relational impedance mismatch does not cause catastrophic performance or integrity loss. Accordingly, object-relational mappers (ORMs) are not only widely used, but also preferred by a majority of developers. In response to our question, "What persistence-related technology do you most enjoy working with?" 58% of respondents answered that they most enjoy working with ORMs. Of these, 70% specifically enjoyed working with Hibernate -  probably a function of both Hibernate's maturity and also our respondents' heavy focus on Java. Although the tail of most-enjoyed data persistence technologies was quite long (26 distinct technologies), Spring Data emerged as the most popular comprehensive data access framework by far (16%).

Reasons Developers Enjoy Working with a Data Persistence Technology
Just under two-thirds of all respondents who named the persistence-related technologies they enjoy working with also specified why they enjoyed working with those technologies. Grounded-theoretic "bucketing" analysis yielded seven (somewhat overlapping) reasons to enjoy a persistence technology (listed in order of popularity): ease of use, simplicity, adherence to standards, familiarity, performance, high level of control, and scalability. The most popular reason by far was ease of use (34%), followed by simplicity in distant second (21%). The top four reasons relate more directly to developer experience than to outcomes (such as performance and scalability), as the wording of the question ("enjoy") indicated. Additional research is required to determine how developer experience relates to persistence-related technology selection, especially because many less-familiar (NoSQL) technologies are optimized for scalability and general performance for certain use cases.

Handling Scale: Data Is Partitioned as Frequently as It Is Not, But This Is Often Successfully Made Invisible to Developers
Modern storage engines, across all storage models, are highly optimized for current hardware, access patterns, and network performance. Theoretically massive inefficiencies of the relational storage model sometimes dominate the advantages offered by a higher degree of maturity among RDBMSes, although newer engines store data in structures that are less narrowly tuned to read-heavy loads using slow (spinning) physical media than (for example) B+ trees. But as Big Data strategies aggressively drive data storage and processing needs, data scale becomes increasingly difficult to manage.

To keep performance and availability high, data is often partitioned on physical and logical lines. Among our survey respondents, 38% partition data in some way (vertical, horizontal, or functional) vs. 40% who do not - a difference within the survey's margin of error (5%). Two research follow-ups would prove interesting: first, what specific data volumes (or velocities), application requirements, and infrastructure constraints drive what kinds of partitioning; and second, which storage models are more likely to require partitioning (although application constraints presumably affect both choice of storage model and partition size/need). It would appear, however, that distributed data techniques designed to manage CAP trade-offs are often effective: 22% of respondents - most of whom are developers and not DBAs -  were not even aware of whether or not their databases were partitioned - a sign that, for nearly a quarter of developers, physical splitting of data had no visible impact on their development work.

For more information on Database and Data Persistence Tools and Techniques, please visit: https://dzone.com/guides/data-persistence-2

More Stories By John Esposito

John Esposito is Editor-in-Chief at DZone, having recently finished a doctoral program in Classics from the University of North Carolina. In a previous life he was a VBA and Force.com developer, DBA, and network administrator. John enjoys playing piano and looking at diagrams, and raises two cats with his wife, Sarah.

IoT & Smart Cities Stories
Early Bird Registration Discount Expires on August 31, 2018 Conference Registration Link ▸ HERE. Pick from all 200 sessions in all 10 tracks, plus 22 Keynotes & General Sessions! Lunch is served two days. EXPIRES AUGUST 31, 2018. Ticket prices: ($1,295-Aug 31) ($1,495-Oct 31) ($1,995-Nov 12) ($2,500-Walk-in)
According to Forrester Research, every business will become either a digital predator or digital prey by 2020. To avoid demise, organizations must rapidly create new sources of value in their end-to-end customer experiences. True digital predators also must break down information and process silos and extend digital transformation initiatives to empower employees with the digital resources needed to win, serve, and retain customers.
Business professionals no longer wonder if they'll migrate to the cloud; it's now a matter of when. The cloud environment has proved to be a major force in transitioning to an agile business model that enables quick decisions and fast implementation that solidify customer relationships. And when the cloud is combined with the power of cognitive computing, it drives innovation and transformation that achieves astounding competitive advantage.
Machine learning has taken residence at our cities' cores and now we can finally have "smart cities." Cities are a collection of buildings made to provide the structure and safety necessary for people to function, create and survive. Buildings are a pool of ever-changing performance data from large automated systems such as heating and cooling to the people that live and work within them. Through machine learning, buildings can optimize performance, reduce costs, and improve occupant comfort by ...
René Bostic is the Technical VP of the IBM Cloud Unit in North America. Enjoying her career with IBM during the modern millennial technological era, she is an expert in cloud computing, DevOps and emerging cloud technologies such as Blockchain. Her strengths and core competencies include a proven record of accomplishments in consensus building at all levels to assess, plan, and implement enterprise and cloud computing solutions. René is a member of the Society of Women Engineers (SWE) and a m...
IoT is rapidly becoming mainstream as more and more investments are made into the platforms and technology. As this movement continues to expand and gain momentum it creates a massive wall of noise that can be difficult to sift through. Unfortunately, this inevitably makes IoT less approachable for people to get started with and can hamper efforts to integrate this key technology into your own portfolio. There are so many connected products already in place today with many hundreds more on the h...
Digital Transformation: Preparing Cloud & IoT Security for the Age of Artificial Intelligence. As automation and artificial intelligence (AI) power solution development and delivery, many businesses need to build backend cloud capabilities. Well-poised organizations, marketing smart devices with AI and BlockChain capabilities prepare to refine compliance and regulatory capabilities in 2018. Volumes of health, financial, technical and privacy data, along with tightening compliance requirements by...
Charles Araujo is an industry analyst, internationally recognized authority on the Digital Enterprise and author of The Quantum Age of IT: Why Everything You Know About IT is About to Change. As Principal Analyst with Intellyx, he writes, speaks and advises organizations on how to navigate through this time of disruption. He is also the founder of The Institute for Digital Transformation and a sought after keynote speaker. He has been a regular contributor to both InformationWeek and CIO Insight...
Digital Transformation is much more than a buzzword. The radical shift to digital mechanisms for almost every process is evident across all industries and verticals. This is often especially true in financial services, where the legacy environment is many times unable to keep up with the rapidly shifting demands of the consumer. The constant pressure to provide complete, omnichannel delivery of customer-facing solutions to meet both regulatory and customer demands is putting enormous pressure on...
Andrew Keys is Co-Founder of ConsenSys Enterprise. He comes to ConsenSys Enterprise with capital markets, technology and entrepreneurial experience. Previously, he worked for UBS investment bank in equities analysis. Later, he was responsible for the creation and distribution of life settlement products to hedge funds and investment banks. After, he co-founded a revenue cycle management company where he learned about Bitcoin and eventually Ethereal. Andrew's role at ConsenSys Enterprise is a mul...