Keeping Financial RAG Applications Always On with DxOperator on OpenShift

Overview

Financial research teams, portfolio managers and strategy analysts increasingly rely on Retrieval-Augmented Generation (RAG) to query earnings reports, investor presentations, analyst research, market data and internal market data using natural language.

However, a production RAG application is only as resilient as its data layer. If the database containing the source content, vector embeddings, and metadata becomes unavailable, the Large Language Model (LLM) can no longer retrieve the trusted context required to provide grounded, accurate answers.

DxOperator by DH2i solves this by automating the deployment and lifecycle management of highly available SQL Server Availability Groups on Red Hat OpenShift. While DxOperator leverages Kubernetes custom resources for deployment, DxEnterprise provides the SQL Server-aware monitoring, clustering, and automatic failover required for mission-critical workloads.

Together, Red Hat OpenShift, SQL Server 2025, DxOperator and DxEnterprise provide a resilient foundation for production RAG applications that demand uninterrupted access to trusted financial data.

Continuously Available RAG Retrieval

DxOperator and DxEnterprise ensure the RAG data layer remains accessible by deploying multiple SQL Server replicas within an Availability Group and coordinating seamless failover. Applications connect through a stable SQL Server endpoint rather than a specific pod; following a failover, the application reconnects to the new primary replica automatically, eliminating the need for manual configuration changes.

Database-Aware Automatic Failover

Unlike standard container orchestration, DxEnterprise provides database-aware monitoring. It tracks SQL Server health, Availability Group states, replica connectivity, synchronization, and cluster quorum. When a primary replica fails, DxEnterprise coordinates an automatic failover to an eligible secondary replica, protecting the RAG retrieval layer from outages that basic Kubernetes health checks cannot resolve.

Resilient AI Across Hybrid Environments

DxOperator streamlines SQL Server Availability Groups on OpenShift, while DxEnterprise extends high availability across Windows, Linux, and Kubernetes environments. This hybrid flexibility allows organizations to migrate to containerized AI services gradually while maintaining connections to legacy SQL Server infrastructure, enabling robust disaster recovery across clusters, sites, and platforms.

Get started with OpenShift

A container platform to build, modernize, and deploy applications at scale.

Try it

Deployment options

Resources

FAQs

What is this AI use case for?

This use case demonstrates how to build a resilient financial RAG application on Red Hat OpenShift, using SQL Server 2025 as the vector and structured data layer. DxOperator and DxEnterprise ensure the SQL Server environment remains highly available so the RAG application can consistently access its source content and embeddings.

What role does DxOperator play in the AI application?

DxOperator does not provide the LLM or the embedding model. Instead, it manages the "engine room"-automating the deployment and lifecycle of the highly available SQL Server Availability Group. DxEnterprise adds the intelligence layer, monitoring the environment and coordinating failovers to prevent downtime.

Why does a RAG application need database-aware high availability?

Standard Kubernetes orchestration can restart a crashed container, but it cannot determine which SQL Server replica holds the most recent data or is eligible to become the new primary. Database-aware HA analyzes instance health, synchronization state, and quorum before failing over, ensuring data integrity and application uptime.

Does the LLM need to run in the same OpenShift cluster?

No. The RAG application can utilize a model deployed on Red Hat OpenShift, Red Hat OpenShift AI, or an external API endpoint. The placement of the model depends on the organization's security and governance needs; regardless of the model's location, DxOperator ensures the supporting data layer remains available.

Keeping Financial RAG Applications Always On with DxOperator on OpenShift

Get started with OpenShift

Get started

Platforms

Products & services

Try, buy, sell

Help

About Red Hat Ecosystem Catalog

Red Hat legal and privacy links

Red Hat legal and privacy links

Keeping Financial RAG Applications Always On with DxOperator on OpenShift

Get started with OpenShift

Get started