Power solution architecture

The architecture for deploying IBM watsonx.data on IBM Power servers by using Red Hat OpenShift is designed to provide high availability, scalability, and efficient resource utilization. This is achieved by using virtualization and container orchestration technologies, helping ensure a resilient and optimized environment for data workloads.

Integrated solution architecture

This integrated solution, as shown in the following figure, consists of watsonx.data deployed on Red Hat OpenShift hosted on three IBM Power servers.

Figure 1. IBM Power architecture

Each Power server is partitioned into multiple LPARs (logical partitions) to provide virtualized, isolated environments that can act as separate servers within a single physical machine.

Red Hat OpenShift is hosted across all three Power servers, where each server has one primary node and one or more compute nodes. This multi-server setup provides key benefits such as high availability and redundancy, scalability, and flexible resource management.

Each Power server hosts an OpenShift Data Foundation (ODF) node to provide S3-compatible object storage for watsonx.data. Additionally, each server includes an infrastructure node that handles cluster services such as routing, registry, and monitoring. Finally, each Power server has a Virtual I/O Server (VIOS) that powers the system's virtualization and I/O functions.

Solution component architectures

IBM watsonx.data on Power inherits the enterprise-grade characteristics of the Power platform, delivering unmatched reliability and quantum-safe encryption. It also provides seamless integration with existing Power workloads, including Oracle databases, Db2 for i, and SAP HANA.

IBM Power infrastructure

IBM watsonx.data on Power is deployed on IBM Power scale-out servers, providing the compute infrastructure for mission-critical AI and analytics workloads. The solution supports both Power10 and Power11 processor generations, enabling organizations to deploy watsonx.data within their existing Power infrastructure while maintaining data sovereignty and using established operational practices.

IBM watsonx.data on Power runs on Red Hat OpenShift container platform, which supports multiple operating systems including Red Hat Enterprise Linux 8.10 or later, RHEL 9.4 or later, or AIX. The appliance is delivered as a rack with all components mounted, cabled, and tested. It provides all the infrastructure resources that are required to host the Red Hat OpenShift cluster, such as storage nodes, compute nodes, and network switches.

The Power scale-out infrastructure provides a robust foundation for deploying watsonx.data at scale while maintaining the enterprise-grade characteristics required for production AI workloads, including seamless integration with existing Power-based data sources such as Oracle, Db2, and SAP HANA.

Object storage

IBM watsonx.data on Power utilizes S3-compatible object storage as the primary storage layer for the data lakehouse. This separation of compute and storage enables independent scaling of resources, cost-optimized storage tiers, and flexible data placement across on-premises and cloud environments.

Organizations can deploy object storage by using Red Hat OpenShift Data Foundation (ODF). ODF provides container-native storage orchestration with NooBaa for S3-compatible object storage within the Red Hat OpenShift cluster. ODF delivers persistent storage for watsonx.data internal services and can serve as the object storage layer for data lake tables.