Google Cloud and DDN Launch High-Performance AI Data Storage

In today’s AI-driven world, data is everything—but if your data platform can’t keep up with your compute, your innovation stalls. Whether you’re training massive foundation models, running large-scale simulations, or processing complex imaging data, performance bottlenecks in storage often become the limiting factor.

That’s why the general availability of Google Cloud Managed Lustre, powered by DDN EXAScaler^® technology, is such a significant milestone. For the first time, enterprises can now access the power of Lustre – the parallel file system trusted by the world’s most demanding compute environments – as a fully managed, cloud-native service on Google Cloud. This offering brings decades of parallel file system expertise from DDN directly into the Google Cloud ecosystem. This is AI infrastructure built to scale with your ambition.

For IT leaders, it’s a way to accelerate data-intensive workloads while optimizing cost and reducing operational overhead. Managed Lustre combines pay-as-you-go flexibility with enterprise-grade performance and support, helping teams move faster, scale smarter, and bring AI, ML, and HPC projects to market faster – without the overhead.

And that changes everything.

Why AI Infrastructure Needs a New Class of Data Storage

AI workloads are unlike anything we have seen before. They demand more than capacity—they demand throughput, latency, and concurrency at levels most storage systems were never built to deliver.

When training a large language model or running distributed image processing pipelines, your infrastructure needs to serve data fast enough to keep thousands of GPUs or TPUs fed at full throttle. And when it doesn’t? You lose time. You waste compute. You exceed your budget. You compromise outcomes.

Traditional file systems—whether cloud-native or legacy—often struggle to meet this threshold. Managed Lustre doesn’t. It provides:

Up to 1 TiB/s aggregate read throughput
Sub-millisecond latency
Millions of IOPS across thousands of clients
Scalable from 18 TiB to 8+ PiB

This is the kind of performance modern AI workloads demand. This is AI data storage without compromise. Now it’s available in minutes – no hardware procurement, no filesystem tuning, no operational burden – on Google Cloud as a 1P offering though a partnership with DDN. Drastically reduce training times and focus on building models, not managing storage infrastructure

Built for Simplicity. Tuned for Scale.

Lustre has long been the gold standard for high-throughput, low-latency file systems used in scientific computing, advanced manufacturing, and AI research. However, one of the biggest barriers to adopting Lustre has always been its complexity. Setting it up, managing it, scaling it—it’s been the domain of HPC experts and dedicated infrastructure teams.

Managed Lustre changes that. Provision a Lustre file system directly from the Google Cloud Console, use the gcloud CLI, or automate it with Terraform. It’s fully integrated with Google Cloud’s monitoring, IAM, and networking tools. And it’s supported globally by Google and DDN together.

You still get the scale, the performance, and the reliability Lustre is known for. But now, you don’t need to manage any of it. Achieve faster iterations, high throughput for data pipelines, seamless integration with AI/ML frameworks (PyTorch, TensorFlow, JAX) and Vertex AI. Now you can start new AI projects, faster. You can focus on innovation instead of infrastructure management.

“This launch represents a significant milestone in the evolution of cloud AI infrastructure. By combining Google Cloud’s leadership in AI with DDN’s decades of expertise in high-performance computing, we’re delivering a fully managed, enterprise-grade parallel file system—purpose-built for today’s most data-intensive AI workloads.

Google Cloud Managed Lustre, powered by DDN EXAScaler®, brings the scale, performance, and simplicity that AI and HPC teams have been waiting for. Now, customers can unlock the full potential of their compute investments—whether training large language models, running complex simulations, or accelerating genomics pipelines—without having to manage the underlying infrastructure. Its great to see two industry leaders come together to simplify and scale next-generation innovation.”
— Santosh Erram, VP, Hyperscaler & Strategic Partnerships, DDN

Data-Intensive Use Cases, Finally Unlocked

Organizations in financial services, life sciences, media, and research are already realizing transformative gains.

This isn’t just about speed—it’s about unlocking workloads that weren’t previously viable in the cloud. Teams across industries are using Google Managed Lustre powered by DDN EXAScaler^® to accelerate discovery, production, and insight.

“We saw 50% faster model training and 30% cost savings by switching to Managed Lustre. The performance tiering and GCP integration made a huge difference.”
– Early Access Customer, Confidential

GenAI & LLM Training

From data preprocessing to large-scale checkpointing, GenAI workloads benefit directly from Lustre’s ability to deliver massive I/O to thousands of compute cores—without pause, drastically reducing training times.

Scientific Research & Genomics

Sequencing pipelines, molecular simulations, and Cryo-EM imaging require sustained, high-throughput data pipelines—both for compute efficiency and to avoid I/O stalls that waste expensive compute time.

Media & Rendering

In VFX, post-production, and animation workflows, teams work on shared files in high resolution across globally distributed environments. Google Managed Lustre brings the parallel performance of on-prem environments to the cloud, without the complexity.

Financial Modeling

Quantitative finance workloads such as risk modeling, option pricing, and backtesting involve massive datasets and require fast, concurrent access from thousands of compute instances. Google Managed Lustre powered by DDN EXAScaler^® offers POSIX compliance and ensures that simulations complete faster, with more iterations possible in less time—enabling better trading strategies, reduced latency in decision-making, and faster time-to-insight.

Engineering Simulations

Workloads like computational fluid dynamics (CFD), structural analysis, and EDA for semiconductor design rely on parallel compute clusters running tightly coupled jobs. These simulations are I/O intensive and often bottlenecked by shared storage systems. With Google Managed Lustre, engineers get consistent throughput and low-latency access to large datasets, improving job efficiency and accelerating design cycles.

Automotive & Autonomous Systems

From digital twin simulations and ADAS model training to massive sensor data replay, modern automotive development is data-intensive at every step. Google Managed Lustre enables rapid access to large-scale test data, simulation models, and AI training inputs—making it easier for OEMs and suppliers to run parallel development pipelines in the cloud. Whether validating a perception model or processing LiDAR streams, Lustre supports the high-speed workflows that autonomous innovation demands.

And with multiple performance tiers, organizations can align costs with use cases—from bursty, short-term model training to always-on inference serving.

“Enterprises today demand AI infrastructure that combines accelerated computing with high-performance storage solutions to deliver uncompromising speed, seamless scalability and cost efficiency at scale. Google and DDN’s collaboration on Google Cloud Managed Lustre creates a better-together solution uniquely suited to meet these needs. By integrating DDN’s enterprise-grade data platforms and Google’s global cloud capabilities, organizations can readily access vast amounts of data and unlock the full potential of AI with the NVIDIA AI platform (or NVIDIA accelerated computing platform) on Google Cloud —reducing time-to-insight, maximizing GPU utilization, and lowering total cost of ownership.”
– Dave Salvator, Director of Accelerated Computing Products, NVIDIA

The Power Behind the Platform

Managed Lustre is the result of a multi-year collaboration between Google Cloud and DDN. At its core is DDN’s EXAScaler^®, the industry’s most trusted Lustre distribution—used in research labs, national supercomputing centers, and enterprise HPC environments worldwide. NVIDIA exclusively uses DDN EXAScaler^® in their internal clusters.

That technology is now running as a first-party service in Google Cloud, designed to be as easy to consume as any other storage class, but with the parallelism, throughput, and tuning that DDN has spent decades perfecting.

“By combining the scale of GCP with the performance of DDN EXAScaler^®, we’re unlocking new levels of throughput for customers training models in days, not weeks.”
– Sameet Agarwal, VP Engineering, Google Cloud

From On-Prem to Hybrid Cloud, Without Compromise

For organizations already running on-prem Lustre environments such as DDN EXAScaler^®, Google Cloud Managed Lustre offers a clear path to elastic, cloud-based hybrid expansion. Whether you’re looking to burst into the cloud for peak workloads or gradually shift more operations off-prem, Managed Lustre delivers the flexibility to grow without disruption:

Extend existing workflows to the cloud without rearchitecting
Eliminate the need for additional on-prem hardware
Shift from CapEx-heavy infrastructure to predictable, usage-based OpEx
Scale storage performance and capacity elastically to meet changing demands
Access the latest DDN and Google Cloud features with no manual upgrades

If you’re already running Lustre on-premises—or managing a parallel file system manually—Managed Lustre offers a direct migration path. You get the same file system semantics, POSIX compliance, and compatibility with standard Linux distributions (including Rocky, Ubuntu, and Debian).

But now it’s managed. Elastic. Integrated. And designed to connect directly with other Google Cloud services—like GKE, Cloud Storage, Vertex AI, TPUs, and more. Simplify provisioning via GCP tools, Terraform support, automated upgrades, integrated monitoring, robust security controls, and POSIX compliance.

With native integration to Google Cloud Storage (GCS) and high-speed bulk transfer tools, moving petabyte-scale datasets into and out of the cloud becomes fast, efficient, and cost-effective—making hybrid deployment not just possible, but practical.

Because the future of storage isn’t about capacity. It’s about capability

Data isn’t just growing—it’s accelerating. And as workloads evolve, infrastructure needs to evolve with them. Google Cloud Managed Lustre, powered by DDN EXAScaler^®, is a foundational step in making AI, HPC, and data-intensive computing faster, simpler, and more accessible.

Because the future isn’t just about storing data—it’s about using it, at speed, and at scale.

Learn More and Get Started

The best way to understand what Managed Lustre can do is to try it. Provision a Managed Lustre instance in the Google Cloud Console in minutes and start testing it against your most I/O-intensive workloads.