Empowering data science teams to streamline model development lifecycles, eliminate dataset versioning bottlenecks, and cut GPU overhead on Google Cloud Platform.
Data-driven enterprises are racing to scale artificial intelligence and machine learning initiatives, but they frequently encounter a common obstacle: data infrastructure bottlenecks. Moving multi-terabyte datasets, managing version control across complex ML pipelines, and coordinating infrastructure between data engineering and AI teams can stall innovation and lead to idle, expensive GPU resources.
To eliminate these barriers, we are excited to announce the General Availability (GA) of the integration between Google Cloud NetApp Volumes (GCNV) and Domino Data Lab on Google Cloud Platform.
By combining Domino's Enterprise AI Platform running on Google Kubernetes Engine (GKE) with the high-performance, intelligent data management of Google Cloud NetApp Volumes, organizations can now deliver self-service data access to AI practitioners while ensuring enterprise-grade governance and efficiency.
The Enabler: GCNV in ONTAP-Mode
The data plane foundation behind this integration is the ONTAP-Mode service level in Google Cloud NetApp Volumes. ONTAP-Mode delivers the full control and feature depth of NetApp ONTAP directly within Google Cloud—fully native and managed.
Why ONTAP-Mode is Essential for Domino Data Lab?
- Native NetApp Trident CSI Orchestration: ONTAP-Mode seamlessly integrates with NetApp Trident CSI and allows deeper SVM level access to ONTAP primitives that are essential for Domino Data Lab running on GKE to dynamically provision, mount, snapshot, and clone persistent volumes directly via Kubernetes primitives.
- Uncompromising Performance at Scale: High-throughput AI workloads require massive performance. GCNV in ONTAP-Mode delivers customized throughput and IOPS scaling—capable of delivering up to 3 million IOPS and 200 GiB/s throughput required for model development and parallel data ingestion.
- Unified REST API Control: Storage administration and MLOps automation can interact directly with ONTAP REST APIs, bridging cloud-native orchestration with enterprise storage operations.
Unblocking Strategic Customer Use-Cases
This joint solution addresses the operational friction points faced by enterprise AI teams across medical, life sciences, engineering, and financial domains.
Scaling Large-Scale Video and Image Datasets for Model Development
Staging and versioning multi-modal data present major obstacles for Healthcare and Life Sciences organizations that must ingest hundreds of TiBs of visual datasets for model development.
- The Challenge: Traditional snapshot copies and full file transfers are painfully slow, and delay time-to-insight.
- The Solution: Through NetApp Volumes and Domino Data Lab, teams can instantly version multi-terabyte repositories using Snapshots. Data scientists can subsequently branch datasets into new volumes using Snapshots for parallel experiment runs in a matter of minutes.
Distributed GPU access & Cost Optimization
High-performance AI clusters require continuous, low-latency file streaming to prevent GPU starvation during model development, fine-tuning, and computer vision tasks.
- The Challenge: Slow storage I/O starves GPUs, forcing expensive compute nodes to sit idle waiting for data.
- The Solution: NetApp Volumes delivers consistent high throughput with support for QoS rate limits at a volume level allowing performance to be directed to datasets as needed eliminating the need for over-provisioning of resources to meet desired performance in the presence of noisy neighbors while Domino auto-scales and auto-shuts-down GPS workspaces so compute runs only when needed, complementing GCNV’s volume level QoS by eliminating idle GPU spend on top of eliminating I/O-driven GPU starvation.
Self-Service MLOps & Cross-Project Data Sharing
Data science workflows thrive on speed and experimentation, but waiting for IT provisioning halts momentum.
- The Challenge: Data scientists frequently rely on DevOps tickets to provision storage, set up access permissions, or replicate datasets across project environments.
- The Solution: Domino Volumes for NetApp ONTAP enables data scientists to provision, attach, and share persistent ONTAP-backed datasets through NetApp Volumes directly within the Domino UI. Teams can collaborate seamlessly across projects using shared, up-to-date volumes without manual data duplication or governance risks.
How ONTAP’s Deep Data Management Stack Powers AI on Google Cloud Platform
The integration between GCNV and Domino Data Lab is powered by NetApp ONTAP’s enterprise data management capabilities, bringing decades of data handling expertise directly into Google Cloud.
- Instantaneous Cloning (FlexClones): Create near-instant copies of datasets for testing, fine-tuning, and multi-developer experimentation.
- Automated Cold Data Tiering: Automatically migrate inactive or historical AI datasets from high-performance storage pools based on activity thresholds, while maintaining filesystem metadata on the hot tier in GCNV.
- NAS and S3 Protocol Duality: Access the same underlying dataset concurrently via traditional NAS file paths and S3-compatible object storage APIs, enabling legacy applications and modern AI frameworks to collaborate on unified data without transformation.
- Hybrid Cloud Mobility: Seamlessly replicate or cache on-premises enterprise datasets up to Google Cloud NetApp Volumes using SnapMirror replication or FlexCache, establishing a global unified data fabric for AI pipelines.
Feature & Business Impact Matrix
Capability | Technical Mechanism | Impact on AI & Data Science Teams |
|---|
Rapid Dataset Versioning | NetApp Snapshot™ & FlexClone® | Replicate multi-terabyte datasets in seconds |
Maximized GPU Utilization | Custom-performance up to 200 GiB/s throughput | Eliminates I/O bottlenecks; while reducing data processing time leading to reduction in GPU runtime cost. |
Self-Service Storage | Domino Volumes UI + Trident CSI | Data scientists provision storage directly inside Domino without IT intervention. |
Cost Optimization | Dynamic Auto-tiering | Reduces overall TCO by auto-archiving cold datasets to lower-cost storage tiers. |
Hybrid Cloud Connectivity | ONTAP SnapMirror® & FlexCache® | Effortlessly stream on-premises datasets into Google Cloud for cloud-native AI workflows. |
Get Started Today
The integration of Google Cloud NetApp Volumes with Domino Data Lab bridges the gap between enterprise data governance and cutting-edge MLOps. By eliminating data transport delays, providing instant dataset versioning, and accelerating model execution, this joint solution empowers data science teams to convert raw enterprise data into AI breakthroughs.
