The HPE Alletra file and block storage portfolio has gained object storage with the Alletra MP X10000 for large-scale unstructured data lakes and digital repositories.
It is designed for exabyte scale and achieves up to 6x faster performance versus the competition, which HPE says are “two of the leading vendors delivering object storage.” That might mean AWS with S3 and Azure with its Blob offering.
HPE storage product marketer Tim Desai writes: “We are excited to introduce the HPE Alletra Storage MP X10000, a unique object storage solution designed to transform the way you handle data-intensive workloads and stay ahead of the curve.”
The X10000 uses Alletra Storage MP base hardware – scale-out storage nodes with a disaggregated shared everything (DASE) architecture. This features ProLiant server chassis with all-flash NVMe fabric-connected local storage and Aruba switches. HPE has OEM’d VAST Data software for its subscription-style GreenLake File Storage offering, which also uses Alletra MP hardware. The GreenLake for Block Storage offering has been rebranded as the Alletra Storage MP B10000 (B for block). We would not be surprised to see an Alletra Storage MP F10000 brand appear for GreenLake File Storage in the interest of consistency.
Desai says: “The X10000 is not just a storage solution. It’s a game-changer for enterprises looking to optimize their data management and drive business value.”
What makes it a game-changer? HPE said that, in addition to its performance and scalability, the X10000 features the GreenLake cloud operational experience with a common Alletra MP platform “that can be configured for different software-defined storage personas and use cases” – block, file, and now object. Desai said: “We are delivering on our vision to provide multiple storage services with multi-protocol support on a single disaggregated cloud managed platform.”
The X10000 has container-native software with Kubernetes-based orchestration, erasure coding, and inline data reduction. It has native AWS S3/S3a API support and incorporates a log-structured key value store, “optimizing the use of flash media for superior performance.” We are told: “There is no front-end caching or data movement between media. Each node adds a proportionate amount of performance to the cluster, enabling seamless scaling for concurrent users and client nodes.”
There is up to 20x data reduction, likely for backup data, and streamlined integration with any backup product; Commvault and Veeam initially with more in qualification.
The entry level is a three-node system and it can grow to hundreds of nodes. Capacity expansion is seamless, requiring minimal data movement, and upgrades can be performed non-disruptively. Performance and capacity can be rebalanced during expansions with minimal impact, eliminating the need for costly data transfers.
The X10000 “allows for future inline data services, bringing compute closer to the data.”
Management via HPE’s Data Services Cloud Console provides centralized monitoring, with a single UI, protection, self-provisioning, and proactive support for global scale infrastructure including Alletra Storage MP B10000, X10000, and GreenLake for File Storage.
By writing its own object storage software from the ground up, HPE appears to have rejected the option of using object storage from its two existing partners, Cloudian and Scality, either through licensing or acquisition. These two suppliers are now competing with in-house object storage from DataCore, Dell, DDN, HPE, Hitachi Vantara, IBM, Infinidat, NetApp, Pure Storage, and Quantum. Lenovo is the last main system vendor with no in-house object storage.
HPE says it’s collaborating with Nvidia to enable a direct data path for direct memory access (DMA) transfers between GPU memory, system memory, and the X10000, which is critical for AI applications. In other words, GPUDirect-like S3 over RDMA, which is already supported by Cloudian and MinIO. We think S3-over-RDMA will become table stakes for all object storage vendors.
Check out an Alletra Storage X10000 brochure here and data protection brochure here.
Bootnote
HPE also announced VM Essentials, a unified VM management facility for virtualized workloads across hybrid environments. It integrates existing virtualized workloads with the new HPE VME hypervisor. VM Essentials supports the main storage protocols, distributed workload placement, high availability, live migration, and integrated data protection. HPE claims that, by using GreenLake cloud and VM Essentials, enterprises can save up to 5x on TCO.
