Data Gravity for IBM Z: Bring Applications to the Data

Introduction – Why Data Gravity Matters

In physics, gravity is determined by mass—the greater the mass, the stronger the pull. Data behaves in a very similar way: as the volume of data increases within a system, it naturally attracts applications, services, and processing to itself. This principle, known as Data Gravity, was first introduced by Dave McCrory in 2010.

In today’s hybrid cloud era, understanding data gravity is essential. Enterprises generate massive amounts of data on core systems like IBM Z and must decide whether to keep it there or replicate it elsewhere. These decisions directly impact performance, cost, agility, scalability, and data access latency.

Image1: Introduction data gravity

Understanding Data Gravity Parameters

Data gravity is driven by multiple factors, not just data volume:

Image 2:  Understanding data gravity patterns

• Data Mass – The larger the dataset, the stronger the pull to keep processing near it.
• Application Mass – Complex applications tightly coupled with data increase gravity even further.
• Acceptable Latency – If workloads tolerate delays, data may be distributed; if low latency is critical, data stays close to applications.
• Frequency of Access – The more often data is needed, the stronger the pull to centralize processing around it.

In short: Higher data and application mass = stronger data gravity; higher acceptable latency = weaker data gravity.

The Cost of Fighting Data Gravity

Enterprises often try to “fight” data gravity by moving or copying data to external platforms for analytics, AI, or reporting. While this may solve short-term needs, it introduces several challenges:

• Security risks - duplicating sensitive data increases exposure.
• High costs - continuous replication and storage can be expensive.
• Data consistency - maintaining accuracy across multiple platforms is difficult.
• Governance challenges - ensuring compliance across environments becomes complex.
• Operational complexity- managing multiple copies slows innovation.

Instead of fighting gravity, the smarter approach is to bring applications to the data.

Image 3: Cost of fighting data gravity

Data Gravity and IBM Z

For many large enterprises, IBM Z remains the system of record, powering mission-critical transactions across industries like banking, insurance, and retail. Transactional data is created on the mainframe and moving it elsewhere simply to process it goes against the principle of data gravity.

Supporting data gravity on IBM Z means enabling new workloads—analytics, APIs, AI—without breaking the close connection between applications and their data.

Future-Ready Approaches to Supporting Data Gravity

Enterprises today don’t just need to respect data gravity they also want agility, simplicity, and support for open standards. That means reimagining how mainframe data is consumed:

• Data Virtualization – Access relational and non-relational IBM Z data with SQL without physically moving it.
• In-Memory Caching – Create caches of frequently accessed mainframe data to enable low-latency access across digital channels.
• Analytical Offload Elimination – Extend Db2 for z/OS beyond transactional workloads to run analytics directly on the platform.
• AI Closer to Data – Bring AI closer to the systems where business-critical data resides, enabling real-time scoring and inferencing without moving sensitive information, and leverage Telum2 and SPYRE accelerator cards to enhance AI and analytics performance on IBM Z.

IBM Solutions That Support Data Gravity

IBM offers a portfolio of solutions that help enterprises respect data gravity while modernizing how data is accessed and consumed on IBM Z:

Image 4: IBM solutions supporting data gravity

• IBM Data Virtualization Manager for z/OS (DVM) – Virtualizes mainframe data, allowing both relational and non-relational sources (VSAM, IMS, Db2, etc.) to be accessed through SQL or APIs without copying the data.
• IBM Z Digital Integration Hub (ZDIH) – Provides in-memory caching of frequently accessed mainframe data, enabling low-latency access for digital applications across channels.
• IBM Db2 Analytics Accelerator – Extends Db2 for z/OS to support complex analytical workloads directly, reducing the need for offloading and keeping analytics close to the data.
• Machine Learning for IBM z/OS (MLz) with Hardware Acceleration – Enables enterprises to build, deploy, and operationalize ML models directly on IBM Z, making real-time predictions within mainframe applications. With the AI inferencing capabilities of Telum2 and Spyre acceleration for LLMs and decoder models, enterprises can securely run cutting-edge AI where their data resides.

Conclusion – Turning Data Gravity into Advantage

Data Gravity is a strategic reality that enterprises can leverage for innovation. By keeping processing close to where data originates, enterprises reduce risk, cut costs, and unlock new possibilities. For organizations running on IBM Z, solutions like DVM, ZDIH, Db2 Analytics Accelerator, and ML for z/OS ensure that you don’t fight data gravity—you leverage it to power innovation across hybrid cloud.

By leveraging data gravity, organizations can accelerate decision-making, enable real-time insights, and maximize the value of their mainframe investments, all while maintaining security and governance across hybrid environments.

References

1. Dave McCrory – Data Gravity Concept

1. https://datagravitas.com/2010/12/07/data-gravity-in-the-clouds/
2. https://datagravitas.com/2011/04/02/defying-data-gravity/

2. IBM Z and Hybrid Cloud Data Solutions

1. IBM Data Virtualization Manager for z/OS (DVM) overview: https://www.ibm.com/products/data-virtualization-manager-for-zos
2. IBM Z Digital Integration Hub (ZDIH): https://www.ibm.com/products/z-digital-integration-hub
3. IBM Db2 Analytics Accelerator: https://www.ibm.com/products/db2-analytics-accelerator
4. Machine Learning for IBM z/OS (MLz): https://www.ibm.com/products/machine-learning-for-zos

3. IBM Telum2 Processor and SPYRE Accelerator Cards

1. https://www.ibm.com/products/z/telum
2. https://www.ibm.com/new/announcements/telum-ii

About the Author

Vishal Gupta is a Senior Solution Architect with the IBM Z Ecosystem at the India Systems Development Lab (ISDL), IBM. He works with Global System Integrators and clients to build solutions across Hybrid Cloud data integration patterns, including real-time Z data access, mainframe data virtualization, synchronization, and application integration. With over 18 years of IT experience in IBM Z, his expertise spans application programming and solution architecting.