Hewlett Packard Enterprise launches first scale-out disaggregated block storage

Hewlett Packard Enterprise (NYSE: HPE) announced Release 3 of HPE GreenLake for Block Storage, based on HPE Alletra Storage MP – the industry's first scale-out disaggregated block storage that enables enterprises to scale capacity and performance independently. Powered by a shared-everything storage architecture, and managed via the HPE GreenLake Cloud Platform, this unique block storage offering brings an on-premises cloud experience, efficient scale, and extreme resiliency and performance to modern mission-critical workloads.

“Optimizing data management is paramount for today’s modern businesses that give precedence to a data-centric approach,” said Kamal Kashyap, Director, India – Storage Business Unit, Hewlett Packard Enterprise India. “The HPE GreenLake for Block Storage MP Release 3 is a significant leap forward, introducing advanced AI-driven performance reporting and analytics, elevating troubleshooting capabilities and offering enriched insights. With this release, we strive to not just broaden the scope of HPE GreenLake for Block Storage offerings but also empower our customers to modernize storage management and address the growing requirements of today’s enterprise data management.”

Release 3 includes new multi-node switch models that offer twice the performance and two-and-a-half times the capacity of their switchless predecessors. The release unlocks the full power of NVMe for Ethernet networking with support for NVMe-oF/TCP, complementing the previously available Fibre Channel, NVMe-oF/FC, and iSCSi connectivity options. Management via the HPE GreenLake cloud platform has also been enhanced with advanced AI-based performance reporting and analytics. A new warranty program for quadruple data compression also ensures greater cost efficiency. As with previous releases, HPE offers a 100% data availability guarantee.

New Storage Architecture

Legacy storage architectures typically consist of two or four high availability (HA) controller nodes, each linked together in pairs. These pairs of nodes are connected to dedicated drives via a midplane in a single hardware enclosure. The midplane is often a limiting factor and determines the maximum performance of the system. To scale performance, enterprises need to purchase a new system with another pair of HA controller nodes, even if they don't need the extra capacity. Beyond creating an “orphan capacity” issue, this approach creates silos that are complex to manage and maintain.

In HPE GreenLake for Block Storage MP Release 3, however, the midplane is flexible and provides more than enough I/O capacity. This allows enterprises to add additional controllers for more performance and, independently and as needed, additional drives and JBOFs to provide more capacity for applications. All storage SSDs are shared by all controllers over a high-speed network, eliminating silos and compensating for multiple node failures.

Release 3 includes two- and four-controller node switch models with a choice of 16- or 32-core configurations. The systems offer support for up to eight JBOF expansion bays, allowing customers to scale from 15.36 TB to approximately 2.8 PB without interruption. 8 to 24 SSDs are supported per chassis, with the ability to upgrade drives in increments of 2 and JBOFs in increments of 1.

The parallel, multi-node, all-NVMe architecture with active I/O processing across all media, controllers, and host ports ensures consistent, predictable performance and ultra-low latency.

AIOps management via cloud console

HPE GreenLake for Block Storage built on HPE Alletra Storage MP delivers a cloud operational experience powered by industry leading AIOps for infrastructure. This reduces firefighting by predicting and preventing disruptions before they occur. Release 3 offers several improvements in this regard via enhanced and extended AI-based performance reporting and analytics that enhance troubleshooting and insights. This includes, for example, trend analysis of headroom utilisation and the identification of latency hotspots per storage volume, as well as improved identification of workload drifts and resource contention. Capacity reports and efficiency metrics have also been improved, and there are new sustainability metrics, such as electricity consumption trends.

HPE GreenLake for Block Storage MP offers multiple technologies to reduce and aggregate data. These include deduplication with express indexing, an extremely efficient compression algorithm, data packing, hardware-accelerated and reservation-free thin provisioning, and virtual copy. These technologies improve economics without sacrificing performance by making the most of flash capacity and improving flash media endurance. Release 3 comes with a 4:1 data aggregation guarantee, which is calculated as the ratio between usable and effective capacity across all collapsible data.