As organizations scale their AI and analytics efforts, a data architecture that is “up to snuff” is even more critical to business success. Data teams are being asked to deliver faster insights, stronger governance, and AI-ready data, all while keeping costs and complexity in check.
As a result, many are rethinking how they modernize legacy data warehouses, manage hybrid and cloud platforms, improve data engineering productivity, and build reliable foundations for semantic layers and analytics that support better, more robust decision-making.
Experts joined DBTA’s webinar, Modern Data Architecture in Practice: What's Actually Working in 2026, to cut through the hype and share what is actually working in modern data environments today.
AI data is posing a challenge for enterprises, Bassam Chahine, principal consultant at NetApp Instaclustr, explained. Massive datasets are needed for machine learning training. There are real-time inference requirements, a combination of structured and unstructured data throughout the business, and sub-second response times for AI is needed.
Enterprise-Grade AI Infrastructure should consist of:
- GPU-optimized instances for training workloads
- Multi-region replication for global AI apps
- Automated backups for model versioning
- SOC 2, HIPAA, GDPR compliance for AI data
- Monitoring and observability for ML pipelines
Performance should include:
- Horizontal scaling for growing embeddings
- In-memory caching for inference
- Optimized for high-dimensional vectors
Citing the definition of data automation from TDWI, Paul Watson-Gover, senior solutions architect at WhereScape, noted, “Data platform automation is much more than simply automating the development process. It encompasses all of the core processes of data warehousing including design, development, testing, deployment, operations, impact analysis, and change management.”
Today’s risks and challenges include increasing complexity and chaos, Watson-Grover said. This is where WhereScape Data Automation comes in.
The solution eliminates time-consuming manual coding and reduces errors; accelerates project delivery from months and years to days and weeks; lowers maintenance costs and simplifies ongoing management; and facilitates smooth, rapid platform migrations (eg., to cloud) with minimal risk.
According to Watson-Grover, data automation benefits:
- Database administrators
- BI/Database developers
- BI/Data architects
- Executives
- Data analysts and data scientists
Michael Cucchi, chief marketing officer at Hydrolix, said the impossible choices every modern data architect faces include managing global scale with real-time performance.
The approach that works the best is:
- Real-time streaming approach - use micro-partitioning for real-time query results
- Object storage reliability and economics as main DB store - keep data always 'HOT'
- Stateless scaling - decoupled Ingest, query and storage
- Allow for bursts at scale of hundreds PB/day without compute over-provisioning
- Long data retention without performance impact - Partition grooming, merges, and summary tables drive high compression rates
Stream data, index all columns, and utilize real-time stream processing and decoupled storage (object store) to achieve high performance and scale. Design for high-cardinality datasets and independently scalable query pools, Cucchi said.
For the full webinar, featuring a more in-depth discussion, Q&A, and more, you can view an archived version of the webinar here.