Db2 and 50 Years of Relational Database Design

3 min read

Looking back to the beginning of Db2.

In the late 70s and early 80s, a team was assembled at IBM’s lab in San Jose to create Db2, the world’s first enterprise-level relational database, based upon Edgar Codd’s seminal paper and subsequent IBM research. Many could not have imagined the impact that their work would have on the world four decades later.

In that time, Db2 running on the mainframe has become (and remains) the mainstay of most of the largest enterprises across the planet. It provides the foundational data services for essential governmental services and manages the data upon which financial, retail, healthcare, and insurance institutions—and, therefore, all of us—depend on every single day.

In the early days, the primary use case for Db2 was as a data warehouse where the main concern was the reliable storage and retrieval of information. Back then, expectations regarding qualities of service were completely unlike today. As demand grew, over successive releases and versions, the development team has incorporated performance, functional, and availability improvements which have led to the database becoming a viable and then an essential OLTP server, capable of servicing banking and retail applications where millisecond transaction response time is of the utmost importance. To give some idea of the scale, one global financial institution regularly runs 1.5 billion transactions executing 118 billion SQL statements through Db2 on Z daily. Others store over 200 terabytes of production data in Db2 and have measured retrieval of 8 billion data records per second.

To optimize performance and efficiency, it was understood that processing should be brought as close to the data as possible. This meant enhancing Db2 and SQL capability to cope with complexity that previously required application logic, such as referential integrity or trigger processing.

The consumer of Db2 data has also changed over time. Instead of traditional, well-behaved, well-controlled batch applications which ran 9-to-5, Db2 data needed to be made directly available through distributed application connections to customers at ATM machines or logging on to services at any time of the day or night. Coupled with this is increasing regulatory demand regarding data security and accountability.

Integration and innovation

The introduction of parallel sysplex for the mainframe and Db2 data sharing exploitation of it in Db2 V4 represented a step change with respect to data application availability, whereby true 24x7 application access to data can be guaranteed regardless of planned or unplanned events—a solution that remains unique even today.

Data sharing is but one example of the deep relationship the database, mainframe hardware, and operating system share, which ensures that the overall solution continues to meet ever-increasing demands for performance, availability, scalability, security, and resiliency. New GDPS Active/Active solutions provide further support for continuous availability and zero downtime.

Db2 AI for z/OS adds the intelligence to recognize and learn from inefficiencies in SQL performance, then self-tunes and self-corrects access paths with little or no user input. AI is also harnessed in other areas, such as automatically observing and learning application access behavior in order to defend billion-dollar businesses from misbehaved or misconfigured applications.

New online schema capability also allows for application schema changes whilst the application remains online, whereas scalability and performance improvements now support single tables containing up to 280 trillion rows of data and millions of inserts per second.

Hybrid architecture

As the amount of data residing in Db2 continues to grow, so has the importance and desire to fully exploit the information it contains. Instead of taking a one-size-fits-all approach, IBM introduced the IBM Db2 Analytics Accelerator—a unique hybrid solution that delivers massive performance gains for analytical queries against Db2 data, enabling real-time operational decision-making without impacting OLTP workloads and without requiring application changes. As a result, Db2, with its analytics acceleration, is a unique example of enterprise-grade hybrid transactional/analytical processing capability. IBM Db2 for z/OS Data Gate integrates with IBM Cloud Pak for Data and offers an innovative way to provide and synchronize valuable Db2 for z/OS data to be consumed within the AI platform whilst recognizing Db2 for z/OS as the system of record for the world’s largest enterprises.

Setting the stage for tomorrow

While you are reading this, Db2 systems running on IBM mainframes across the world are quietly and securely processing billions of transactions. Db2’s deep integration across the Z stack, exploitation of hardware technology and acceleration, unique hybrid architecture, and embedded AI capability ensures that it remains at the forefront of technological innovation that is necessary to support mission-critical workloads for the world’s largest enterprises long into the future.

Read more about IBM’s rich heritage in driving innovation in data technology.

Be the first to hear about news, product updates, and innovation from IBM Cloud