Cyber recovery at scale: rethinking backup speed and efficiency

The value of business data – regardless of industry – is highlighted by the intense daily battle to keep it safe. While we see rapid advancements in cybersecurity, most companies know that a cyber incident is no longer an if. It’s sadly a case of when.

Whether it's valuable data being deleted from a server or files locked by ransomware, organizations know they need comprehensive, fast backup and restore services to keep their organization running.

Businesses often view backups as an insurance policy – one they hope to never invoke. But the digital world we live in is not without risk, and backups that are overly focused on cost optimization rather than recovery time are no longer acceptable.

The requirements have shifted to high-performance recovery that’s both simple and scalable, restoring a minimum viable state for continued organizational operation. This creates a distinction between traditional backup/restore and cyber recovery.

A UK Government survey shows that over four in ten businesses (43%) and around three in ten charities (28%) reported having experienced any kind of cyber security breach or attack in the last 12 months.

Average downtime following a ransomware attack across industries in 2025 was around 24 days — over three weeks of a business not fully operating, impacting revenue, customer confidence and reputation.

Whilst traditional backup platforms are often used and depended on for recovery of individual workloads or data sets, they face a significant challenge when the need is to recover greater quantities of data. These platforms often make use of two technologies – data reduction and HDDs. This combination makes a fantastic pairing for ingesting backup data and reducing overall cost. The tradeoff is lower performance for data restoration.

Recovery following a cyber event is dependent on access to data that’s instantly recoverable. This is where HPE Alletra Storage MP X10000 comes into its own, delivering enterprise performance with data intelligence embedded directly into the storage layer, with the throughput and concurrency needed for ultra-fast recovery.

ITPro spoke to Richard Trembecki, HPE sales architect for file and object storage, about the ways the X10000 can help.

The X10000

At the heart of HPE’s solution lies the X10000, a disaggregated, shared-everything architecture that’s designed to use standard hardware components and enable the independent scaling of compute nodes and flash storage expansion shelves. It gives organizations the flexibility to increase storage capacity, independently from compute nodes, in contrast to traditional paired-node storage systems.

HPE’s internally developed all-flash HPE Alletra MP X10000, which came to market at the end of 2024, offers high-performance, scale-out object and file storage based on the same modular hardware architecture as the company’s block and file-based array, the HPE Alletra Storage MP B10000.

While this platform can act as a high-performance repository for backup and restore workloads, this can be taken to the next level with the introduction of the Data Protection Accelerator Nodes.

Data Protection Accelerator nodes

Data protection accelerator nodes [DPAN] act as a high-performance ingestion and rehydration engine, which processes, deduplicates, and encrypts backup data before sending it to primary storage repositories. The DPAN on the X10000 does all this and more. It aims to redefine how enterprise backup and recovery is delivered by shifting from legacy and capacity-bound backup appliances to a high-performance, scale-out data protection architecture.

When paired with the X10000, the DPAN passes data straight through to the X10000, improving performance by taking the workload of data management off the storage layer. The result is a lean, high-performance, data protection architecture optimized for solving the challenge of restoring data at both scale and at speed.

The DPAN integrates into Commvault, Veeam, and Cohesity using the HPE StoreOnce Catalyst protocol, making use of variable-length chunking to deduplicate data across all backups, clients, and workloads. Leveraging source-side deduplication and compression to achieve data reduction ratios of up to 60:1 — dramatically cutting the underlying storage required to host the data and the volume of data transmitted during backup. Built-in encryption and immutable backup copies add protection against cyber attacks.

Whilst the performance of a single DPAN is excellent, by connecting multiple DPANs with a single X10000, the solution is able to achieve speeds of up to 2.5PB/hr whilst continuing to deliver the value of the Catalyst feature set, including data reduction, immutability, encryption and more.

How does this affect businesses’ recovery point objective (RPO) and recovery time objective (RTO)? According to Trembecki, HPE’s X10000 and DPAN shrink RPO by focusing on RTO.

“That paradigm shift is significant,” he says. “Hybrid media is not suitable for this challenge. We need an all-flash solution to make sure that we can read the data back as quickly as possible. This means that the X10000 is a good fit because of the read performance as well as the write performance of the platform. The disaggregated nature of the X10000 means that we can scale according to what each customer needs – across performance and/or capacity. What we’ve done is taken it a stage further, and we’ve integrated a data protection accelerator node using our StoreOnce technology into the platform.”

HPE’s Catalyst engine excels at deduplicating data, compressing it, and then storing it in a very small footprint. Roughly a 60-to-1 data reduction rate, according to Trembecki.

“So we can take six petabytes worth of full backups and store them on 100 terabytes worth of capacity. We’ve taken that same technology and put it across the front of the X10000, which means that we can continue to really shrink down the footprint of the data to make that performant but expensive NVMe media go further. We can hit some massive performance numbers, 2.5 petabytes an hour of ingest and over 400 terabytes an hour of restore speed,” he explains.

“This represents a step change in a customer's ability to recover quickly from a cyber event.”

The concept here is that backup is an insurance policy you rarely need to call on at scale, according to Trembecki.

“Cyber recovery presents a different challenge that cannot be met by traditional backup,” he says.

“The need for customers to face this challenge is becoming ever more frequent with the increasing attack rate. The combination of the DPAN and X10000 can simplify the solution to that challenge”.

With HPE Alletra Storage MP X10000, you can start with three nodes and expand without disruption, while ensuring storage efficiency and high availability.

Beyond cyber recovery: powering AI and analytics

Beyond cyber recovery, the same architecture that makes the X10000 fast to restore also makes it a strong foundation for AI and analytics workloads.

The platform's Nvidia-certified storage validation confirms it can deliver data to accelerated compute efficiently enough to support faster model training, lower-latency inference, and higher GPU utilization, meaning organizations aren't forced to run a separate high-performance tier just for AI initiatives. With unstructured data growing far faster than most organizations' ability to manage it, having a single scale-out platform that can serve both backup and AI pipelines — without re-architecting or duplicating data — is increasingly the difference between AI projects that stall at the storage layer and ones that scale.

The X10000 also embeds data intelligence directly into the storage layer, simplifying the data pipelines that AI workloads depend on. Rather than routing data through separate cataloguing, indexing or metadata tools before it can be used, intelligence built into storage lets organizations discover, classify and serve the right data at the point it's needed — cutting out the integration overhead that typically slows AI projects down.

That matters increasingly for agentic AI, where agents need fast, reliable access to accurate, well-contextualized data to act autonomously rather than periodic batch extracts. By combining native file and object protocols with the performance of an all-flash, disaggregated design, the platform delivers enhanced, ready-to-use data straight to AI agents and pipelines, turning storage from a passive repository into an active, responsive part of the AI stack.

Ultimately, organizations are laying the foundations to build an architecture that will help them shrink recovery time, not just backup cost, both now and in the future.

Learn more about HPE Alletra Storage MP X10000 and Data Protection Accelerator Node here.