DH2i adds three OpenShift patterns for AI data resilience

Red Hat and DH2i have added three joint use cases to the Red Hat Ecosystem Catalog, all aimed at keeping AI applications connected to SQL Server data across OpenShift and hybrid environments. The announcement pairs DH2i’s DxOperator, DxEnterprise and DxOdyssey products with three different infrastructure problems rather than presenting one general reference architecture.

Availability for RAG data

The first pattern places SQL Server 2025 vector embeddings and metadata behind a financial-research RAG application. DxOperator manages synchronous SQL Server Availability Group replicas on OpenShift, while DxEnterprise monitors database health and automates failover.

The design addresses a practical dependency: a retrieval system cannot ground model responses when its embedding and metadata store is unavailable. Red Hat says synchronous commit waits for transaction logs to be hardened on a secondary replica before completing a transaction. The post attributes failover measured in seconds to DH2i, but does not publish test conditions or independent results, so teams should treat that number as a vendor claim to validate under their own load and failure model.

A staged migration path

The second catalog use case targets public-sector systems that cannot move older SQL Server environments directly into containers. Its “crawl, walk, run” sequence starts by rehosting Windows virtual machines on OpenShift Virtualization, then moves database workloads from Windows to Red Hat Enterprise Linux, and finally places SQL Server Availability Groups in containers managed by DxOperator and DxEnterprise.

That sequence makes the infrastructure boundary explicit. OpenShift can host virtual machines and containers on the same platform while the database architecture changes in stages. The article says DxEnterprise incorporates DxOdyssey’s connectivity capabilities throughout the transition.

Application-scoped hybrid access

The third pattern connects a fraud-investigation AI service on OpenShift to selected data endpoints without opening broad network access. DxOdyssey creates application-level micro-tunnels between authorized workloads and endpoints, an approach the companies position as zero-trust network access for data that must remain distributed.

Red Hat says that layer can sit alongside OpenShift controls including SPIFFE/SPIRE workload identity, network segmentation through Red Hat Advanced Cluster Security and Confidential Containers for encrypted-memory execution.

The new catalog entries are deployment patterns, not announcements of a new OpenShift or DH2i product release. Their significance is architectural: each makes the data plane behind an AI workload—availability, migration state and connectivity scope—a separately managed platform concern. Before adoption, teams should verify SQL Server licensing and support boundaries, failover behavior, recovery-point objectives and the operational consequences of automated network isolation.