By the DDN Team
A Major Leap Forward in AI Cloud Infrastructure
Today at Google Cloud Next 2025 in Las Vegas, we’re celebrating a transformative milestone; the announcement of Google Cloud Managed Lustre, a fully managed, high-performance parallel file system running in Google Cloud. Powered by DDN EXAScaler®, this zonal storage solution supercharges AI/ML workloads by providing a petabyte-scale data service that delivers sub-millisecond latency, millions of IOPS, and TB/s of throughput.
Google Cloud Managed Lustre will enable enterprises and startups around the world to accelerate AI innovation by utilizing one of the fastest AI model training platforms on the market, coupled with 15x quicker speed for checkpointing and model loading. This game-changing solution dramatically reduces time-to-market for AI initiatives, cuts infrastructure management costs by up to 40%, and allows organizations to deploy production AI applications in weeks rather than months, delivering measurable ROI and competitive advantage in today’s AI-driven economy.
Organizations can confidently build on the same proven technology that NVIDIA has trusted exclusively for their internal clusters since 2016, ensuring enterprise-ready performance for the most demanding AI workloads:
For over 25 years, DDN has powered the world’s most demanding data-intensive environments for leaders like NVIDIA and xAI. Now, in collaboration with Google Cloud, we’re bringing that expertise to the cloud (learn more about our collaboration here), DDN is excited to highlight how this Managed Lustre offering can be paired with DDN Infinia – available now on the Google Cloud Marketplace – to deliver a data intelligence platform for end-to-end AI that drives real-world business outcomes.
Together, Managed Lustre and Infinia create a seamless, scalable, and efficient ecosystem that spans pretraining, training, and inference—unlocking the full potential of your data, anywhere, for any application.
Introducing Google Cloud Managed Lustre – Powered by DDN
A Fully Managed Parallel File System for AI/ML
Today at Google Cloud Next 2025 in Las Vegas, we’re thrilled to announce the availability of Google Cloud Managed Lustre — a fully managed, high-performance parallel file system based on DDN EXAScaler®. This zonal storage solution delivers petabyte-scale capacity with sub-millisecond latency, millions of IOPS, and up to 1 TB/s of throughput – tailored to accelerate AI/ML and HPC workloads. Available directly from the Google Cloud console.
Powered by DDN EXAScaler®, this service is tailored for AI/ML workloads, as it delivers exceptional performance without the complexities of self-managed Lustre deployments. With Google Cloud Managed Lustre, organizations can access an enterprise-grade solution, starting small (terabytes) and scaling elastically as their AI initiatives grow. The platform scales reliably and securely with simplified operations and built-in multi-tenancy capabilities, eliminating traditional infrastructure barriers.
Key benefits include:
- Unmatched Performance: Confidently store your training, checkpoint, and model data, while delivering high throughput, sub-millisecond latency, and millions of IOPS across multiple jobs — all while maximizing output. With its full-duplex network utilization, Managed Lustre can fully saturate VMs at 20GB/s and can deliver up to 1TB/s in aggregate throughput. Enterprises can therefore reduce wait times and speed up time-to-insight using AI applications in Google Cloud.
- Fully Managed Simplicity: Google Cloud handles deployment, scaling, and maintenance, freeing your team to focus on leveraging data intelligence for innovation.
- Scalability: Scales from terabytes to petabytes with full POSIX compliance and persistent zonal storage, backed by a pay-as-you-go mode.
- Seamless Integration: Works with Google Cloud services like Compute Engine, Google Kubernetes Engine (GKE), and Google Cloud Storage (GCS) for end-to-end data intelligence workflows. Support for the Cloud Storage bulk import/export API makes it easy to move datasets to and from Cloud Storage.
- Cost Efficiency: Lowers total cost of ownership (TCO) compared to on-premises Lustre, with no need for specialized expertise.
This is the go-to solution for organizations building LLMs, enterprises bursting from on-premises to the cloud, and teams needing secure, high-performance storage. Learn how
Jump Trading leverages DDN for AI-driven financial analysis.
DDN Infinia on Google Cloud Marketplace: End-to-End AI Data Intelligence
Accelerating Pre-training, Inference, and RAG
While Managed Lustre powers training and does the heavy lifting during AI training, checkpointing, and model loading, DDN Infinia complements it with high-performance data preparation, data analytics and inference capabilities. With the recent announcement of DDN Infinia availability in Google Cloud at NVIDIA GTC 2025, Infinia and Managed Lustre can work together as a comprehensive data intelligence platform within Google Cloud.
Pre-training with DDN Infinia
Infinia supports multiple TB/s bandwidth and sub-millisecond latency to enable:
- Ultra-fast data ingest & organization, unifying multimodal data from across the enterprise (core, cloud, edge).
- Metadata rich indexing (millions of tags per object) to minimize data movement and speed results.
- Seamless integration with frameworks like NeMo™, NIMs, TensorFlow, PyTorch, Trino and Spark to simplify AI workflows.
Training with Managed Lustre
Once data is prepared, Managed Lustre takes over, providing the high-performance data platform needed to train large-scale AI models. Its efficient checkpointing speeds development, cuts costs, and improves the reliability of results. Its seamless integration with Google Cloud’s AI ecosystem ensures fast, reliable data access, minimizing idle GPU time and accelerating results within a data intelligence framework.
Inference with DDN Infinia
Post-training, Infinia steps in again for data analytics, real-time inference and retrieval-augmented generation (RAG). Features include:
- RAG-enabled indexing
- 10x lower-latency data access
- Efficient GPU utilization for production-scale deployments
Why This Combination Matters
Together, Managed Lustre and DDN Infinia on Google Cloud form a unified, enterprise-grade AI data intelligence platform that:
- Accelerates Innovation
- 35% faster model training = faster time to market
- Rapid data prep and AI lifecycle management
- Provides Significant Cost Savings
- 10x data reduction with Infinia
- Managed Lustre’s cloud-native pricing eliminates up front infrastructure costs
- Simplifies Operations
- No infrastructure to manage
- Deploy in minutes via Google Cloud Marketplace
- Scales for the Future
- Infinia elastically scales from terabytes to exabytes with secure multi-tenancy
- Managed Lustre expands across Google Cloud regions by July 2025
Tailored Outcomes for Enterprises and Gen AI Startups
For Enterprises Modernizing AI Infrastructure
DDN and Google Cloud eliminate the need for complex refactoring or costly data movement. Enterprises can accelerate AI adoption timelines by 50% and reduce TCO by up to 60% compared to on-prem infrastructure.
- Maximize GPU utilization and reduce AI training costs with ultra-high-throughput storage.
- Integrates into systems without disrupting legacy applications through native SDKs.
- Avoid cloud egress fees by running inference and RAG directly in Google Cloud.
- Go from purchase to production in days with fully managed, pre-integrated solutions.
For Gen AI Startups Scaling Fast
Agility is everything. DDN + Google Cloud enables fast model development and seamless scaling.
- Train and iterate faster with sub-millisecond data access.
- Integrate directly with tools like NeMo, Vertex AI, and PyTorch to simplify development and speed results.
- Scale from TBs to PBs without re-architecting infrastructure.
- Focus on building core Gen AI products—not managing backend infrastructure.
Whether you’re reining in skyrocketing AI training costs or deploying mission-critical Gen AI applications, DDN and Google Cloud deliver speed, simplicity, and real business results.
See It in Action at Google Cloud Next 2025
Catch live demos of Managed Lustre at the Google Cloud booth, showcasing AI workload acceleration with the DDN-powered platform.
🎤 Session to attend: BRK2-025: What’s New With Cloud Storage
Gain insights into how these solutions transform data intelligence workflows across industries.
A New Era of AI Storage and Data Intelligence
With Google Cloud Managed Lustre and DDN Infinia, customers now have access to the most complete, scalable, and high-performance AI data infrastructure available.
Whether you’re building LLMs, scaling generative AI, or enabling autonomous intelligence, DDN and Google Cloud are your foundation for success.
Learn more about our Google Cloud partnership.
Authored by the DDN Team
