HPE extends GreenLake for Block Storage

HPE has expanded its GreenLake for Block Storage offering with an Alletra MP-based high-end system.

It first announced subscription-based GreenLake for Block Storage – with general purpose Alletra 5000, Alletra 6000 business-critical and Alletra 9000 mission-critical tiers and a 100 percent data availability guarantee – in March 2022. A release 2 added Alletra MP support in in April last year, positioned as mission-critical storage with midrange economics, saying it was the storage industry’s first disaggregated scale-out block storage and featured 8-core and 16-core controller chassis. Since then HPE added non-switched 2-node systems supporting from 8 to 32 cores with up to two JBOF expansion shelves. These were targeted at small and mid-market customers. Now it has generally available faster switched systems.

Simon Watkins, senior manager, worldwide product marketing for HPE Storage & Cloud Data Services, blogged: “Our announcement represents the next step in a multi-step journey to consolidate, standardize and simplify our HPE block storage portfolio on a common software and hardware stack – all with a unified cloud management experience provided by the HPE GreenLake platform.”

HPE has upped the performance and capacity limits substantially, claiming it’s reimagining storage with an architecture that delivers scale, performance, and efficiency in a way that legacy storage competitors cannot achieve. 

There are 16-core/32-core 2-node and 4-node switched models, which deliver up to 2x performance boost compared to previous models. They support up to 2.5x more capacity through up to 8JBOF expansion shelves and a 15.36TB to 2.8PB raw capacity range. You can upgrade drives in 2-drive increments and JBOFs in increments of one. Storage capacity and performance can be scaled independently. A new HPE StoreMore Guarantee guarantees 4:1 data compaction.

The network side has added NVMe/TCP support complementing the existing fiber channel, NVMe-of/FC and iSCSI connectivity. The Infosight AIOps system management tool now enables users to see in real time the impact that workloads are placing on storage and adapt their infrastructure to anticipate and avoid failures and optimize performance. 

There is extended AI-based performance reporting and analytics for improved troubleshooting and insights. Analysis topics include headroom utilization trend analysis, top volume hotspots by latency, performance (latency and throughput), workload drift detection, resource contention detection and sustainability metrics, including power consumption trends.

The 2x performance increase is compared to existing 2-node (16-core/32-core) switchless configurations and the 25x capacity increase is compared to the prior release 2 of GreenLake for Block Storage built on Alletra Storage MP. 

HPE tech specs table.

This new Alletra MP system is for mission-critical apps – from large databases to modern cloud native apps to consolidated mixed workloads – at a midrange price point. Example workloads include PostgreSQL and SAP HANA, for which the new system is certified for up to 80 production nodes.

There is unified cloud management through the GreenLake hybrid cloud platform to enable users to manage and monitor their fleet of block storage globally from a single SaaS-based cloud console accessible from any location, on any device. HPE says the system supports on-line, non-disruptive, data-in-place upgrades. 

It claims legacy storage competitors, many with dual-controller designs, are unable to match the combination of disaggregated scale-out block storage of the Alletra MP system, 100 percent data availability guarantee, and the intuitive cloud experience of the GreenLake hybrid cloud platform. Watkins blogs: “It’s a storage architecture in which storage SSDs are shared by all controllers across a high-speed network, so you have no silos and the ability to tolerate multiple node failures.”

Competition

This Alletra MP block storage system gives HPE something to better answer all-flash array product pitches from Dell (PowerStore), IBM (FlashSystem), NetApp (ASA array) and Pure Storage (Flash Array//). However their maximum capacity goes beyond that of HPE’s system, which scales out to 2.9PB raw and 11.6PB effective capacity.

  • Dell’s PowerStore 500T provides up to 4.71PB per appliance and 18.83PB per cluster. This is effective capacity with 4:1 data reduction.
  • IBM’s FlashSystem 9500 offers 4.5PB effective capacity with 3:1 data reduction in a single enclosure and up to 18.4PB effective with two control enclosures and 5:1 reduction.
  • NetApp’s ASA A900 scales out to 351PB effective capacity – way beyond HPE’s system. Its entry-level ASA 150 can orovide 13PB effective from its up to 12 nodes.
  • Pure’s FlashArray//XL170 scales out to 1.48PB raw and 5.5PB effective capacity.

We think Alletra MP scale out limit improvements could be on HPE’s roadmap.

Watch a video of the performance reporting features here. A December 2023 technical deep dive blog by HPE distinguished technologist Dimitris Krekoukias provides more information. He explains: “The initial release of the new expanded HPE GreenLake for Block Storage will be switchless arrays with disks inside the main chassis, and a near future update will bring the larger, switched, fully disaggregated, shared-everything architecture for much larger and more flexible implementations.” That near future update is what has just become generally available.

Bootnote

The data reduction and data compaction technologies to reduce storage capacity utilization include dedupe with express indexing, an “extremely efficient compression algorithm,” data packing, hardware accelerated and reservation-less Thin Provisioning, and Virtual Copy.

HPE’s 4:1data compaction ratio guarantee is calculated as a ratio of usable to effective capacity, in aggregate, across all reducible data. If not, HPE will provide expertise related to data efficiency, additional storage capacity needed to reach the targeted data efficiency, or other compensation, at its own discretion. The guarantee is valid for the term of the support contract.

HPE also offers customized guarantees, valid for a year, for customers requesting a guarantee for a set amount of effective capacity.