Introduction
As enterprises increasingly turn to artificial intelligence (AI) to drive innovation, the underlying storage infrastructure must evolve to keep pace with growing data volumes and performance demands. Recognizing this shift, Pure Storage has unveiled its next-generation FlashArray and FlashBlade product lines, purpose-built to accelerate AI workloads. These new all-flash platforms deliver unprecedented throughput, ultra-low latency, and embedded AI-driven operations, enabling organizations in New Zealand and worldwide to harness the full potential of generative AI, machine learning, analytics and more.
Why AI Demands Next-Gen Storage
1. Explosive Data Growth: Training large language models (LLMs), computer vision systems and other AI applications requires ingesting and processing petabytes of structured and unstructured data. Legacy disk-based arrays and even previous-generation flash arrays struggle to deliver the sustained performance and scalability these workloads demand.
2. Performance at Scale: AI pipelines typically involve massive parallel I/O across hundreds or thousands of GPUs and CPUs. Storage solutions must provide extremely high IOPS, massive bandwidth and consistent low latency to avoid bottlenecks that slow down training or inference.
3. Operational Complexity: AI environments are dynamic. Models need constant retraining, data pipelines evolve rapidly, and infrastructure must adapt without costly downtime. Manual tuning and troubleshooting of storage systems at AI scale is impractical.
Introducing the Next-Gen FlashArray
Pure Storage’s flagship FlashArray family receives a ground-up redesign, delivering up to 5× higher performance and 3× greater capacity compared with prior generations. Key highlights include:
• NVMe-Optimized Architecture
– Full NVMe-oF (over Fabrics) support for both RDMA and TCP transports
– Dedicated hardware accelerators for inline compression and encryption
– End-to-end NVMe queueing for sub-200 microsecond latency
• Multi-Petabyte Scaling
– Up to 24 PB effective capacity in a single chassis with Pure’s industry-leading data reduction
– Non-disruptive scaling from 20 TB to multiple petabytes to match evolving AI data sets
• Embedded AI Operations
– InfoSight AI-driven monitoring and predictive analytics built into the array
– Automatic workload profiling, performance optimization and anomaly detection
– Self-healing capabilities to preemptively address issues before they impact AI pipelines
• Data Mobility and Protection
– Integrated replication, snapshots and instant restore for multi-site AI workflows
– ActiveDR continuous replication for zero-data-loss disaster recovery
– File and object support to unify structured and unstructured AI data
Introducing the Next-Gen FlashBlade
To tackle unstructured data workloads—such as training image, video and genomic AI models—Pure’s FlashBlade line also gets a major upgrade:
• Scale-Out Performance
– Linear scaling to 5 PB raw capacity per system and multiple exabytes across clusters
– Aggregate throughput exceeding 200 GB/s and 10 million IOPS in a single 4U chassis
• Unified File and Object Access
– NFS, SMB and S3 support in one platform, streamlining data pipelines for AI, analytics and backup
– Concurrent high-speed access for both streaming and small-file workloads, ideal for mixed AI datasets
• Built-In Data Services
– Inline deduplication and compression tailored for video, image and large object workloads
– Policy-driven tiering to on-premises or cloud targets for cost-effective long-term retention
– Snapshot-based clones to spin up hundreds of ephemeral AI training environments
• Simplified Management
– Centralized dashboard in Pure1 to provision, monitor and optimize FlashBlade clusters
– Automated firmware and software updates with zero downtime
– Role-based access control and audit logging for secure AI deployments
Embedded AI-Driven Services
Beyond raw hardware upgrades, Pure Storage has enhanced its Pure1 management platform to serve as an “AI-Ops” nerve center:
• Predictive Support: InfoSight’s machine learning models analyze telemetry from thousands of arrays globally, identifying performance trends and potential faults before they occur.
• Automated Tuning: Workload recognition continuously adjusts Quality of Service (QoS) settings, ensuring critical AI jobs receive priority I/O without manual intervention.
• Capacity Forecasting: Pure1 predicts future storage consumption based on historical AI workload growth, allowing IT teams to budget and plan expansions proactively.
• ChatOps Integration: A new AI-powered chat assistant provides natural-language access to storage health, performance metrics and troubleshooting guidance—right within messaging platforms.
Availability and Pricing
The next-generation FlashArray and FlashBlade systems are generally available today in New Zealand through Pure Storage’s network of channel partners and resellers. Pricing is subscription-friendly, with flexible licensing that includes:
• Evergreen™ Storage Subscription: All-inclusive software, support and hardware refreshes every three years.
• Pure as-a-Service: Pay-per-use consumption model to align storage costs with AI workload demands.
• Modular Expansion: Scale compute and capacity independently for optimized TCO.
Organisations can leverage Pure Storage’s professional services and certified AI specialists to plan, deploy and fine-tune their AI infrastructure, ensuring rapid time-to-value.
Conclusion
With AI initiatives moving from pilot projects to mission-critical enterprise programs, storage infrastructure can no longer be an afterthought. Pure Storage’s next-generation FlashArray and FlashBlade platforms deliver the high throughput, low latency and AI-driven operational intelligence required to power large-scale AI and data analytics. By unifying block, file and object storage on all-flash arrays and embedding advanced AI services into every layer, Pure Storage aims to remove storage as a bottleneck, letting organisations focus on extracting insights and driving innovation.
3 Key Takeaways
1. Performance and Scale: Next-gen FlashArray and FlashBlade deliver up to 5× higher performance, multi-petabyte scaling and sub-200 μs latency to support demanding AI workloads.
2. AI-Driven Operations: InfoSight and Pure1 bring predictive analytics, automated tuning and ChatOps integration, radically simplifying management at AI scale.
3. Flexible Consumption: Evergreen subscriptions and Pure as-a-Service provide predictable costs, future hardware refreshes and pay-per-use models tailored to AI projects.
3-Question FAQ
Q1: How do these new arrays differ from Pure Storage’s previous generation?
A1: They feature a full NVMe-optimized architecture with hardware accelerators, scale up to 24 PB in a single FlashArray chassis or 5 PB raw in a FlashBlade, deliver up to 5× performance gains and embed AI-driven support and tuning via InfoSight.
Q2: What types of AI workloads benefit most from the new FlashArray and FlashBlade?
A2: Generative AI model training, machine learning pipelines, high-performance analytics, genomics, video processing and any workload requiring massive throughput, low latency and large-scale unstructured data access.
Q3: When and how can New Zealand organisations purchase these systems?
A3: The next-gen arrays are available immediately through Pure Storage’s NZ channel partners. Organisations can opt for perpetual licensing, Evergreen subscriptions or Pure as-a-Service, and can engage Pure’s professional services for design and deployment support.
