Catalogue · Hewlett Packard Enterprise
HPE Alletra Storage MP X10000
High-end · all-flash · scale-out - Disaggregated all-NVMe scale-out S3 object and NFS file system for AI data pipelines, data lakes and fast backup targets.
| Platform | |
|---|---|
| OEM | Hewlett Packard Enterprise |
| Family | Alletra Storage MP X10000 |
| Class | High-end |
| Architecture | Scale-out cluster |
| Status | Current |
| Released | n/a |
| Operating system | HPE Alletra Storage MP X10000 OS (1.x) |
| Controllers | 16 |
| Nodes | 3 - 16 |
| Cache / memory | n/a |
| Availability | n/a |
| Capacity and media | |
| Media | All-flash |
| Drive types | NVMe SSD, QLC NVMe |
| Drive note | Self-encrypting NVMe SFF SSDs, TLC (3.84-30.72 TB) or QLC (15.36-61.44 TB); all drives in a cluster must be TLC or QLC; 24 SSDs per JBOF |
| Flash drive sizes | 3.84, 7.68, 15.36, 30.72, 61.44 TB |
| HDD sizes | n/a |
| Maximum drives | 384 |
| Controller enclosure | 2U |
| Expansion | 1-16 X10000 JBOF storage shelves (2U, 24 NVMe SSDs, 1,474.56 TB raw max each); optional Data Protection Accelerator nodes (StoreOnce Gen5-based, HPE Catalyst target) and a Data Intelligence node (NVIDIA L40S GPU) - 24 bays, 2U, up to 16 |
| Raw capacity | 23.59 PB |
| Cluster maximum | n/a |
| Effective maximum | 35.11 PB |
| Data reduction | Effective capacity assumes 2:1 compaction (compression) in a triple-parity RAID (20+3+1) layout |
| Protection | Triple-parity RAID (20+3+1) |
| Protection note | Erasure coding; triple-parity (20+3+1) layout is the effective-capacity basis. Bi-directional (active/active) bucket-level replication between X10000 systems; S3 Object Lock and Versioning. 'ib' here means RDMA file access over 100GbE. |
| Performance and connectivity | |
| Maximum IOPS | n/a |
| Maximum throughput | n/a |
| Latency | n/a |
| Conditions | A 2U chassis holds two controller nodes (drive bays unused) or 24 SSDs (JBOF); 3-16 controller nodes and up to 16 JBOFs per system; drives_max = 16 x 24. Per node 2-4 x 100GbE or 4-8 x 10/25GbE host ports (no mixing); host_ports_max = 8 x 16. RDMA for file only on 100 Gb networks. power_w = max power of one 2U controller enclosure (two nodes, no drives/HBAs: 1072 W max, 780 W idle). HPE notes limits will increase with future software. |
| Protocols | S3 object, NFS, InfiniBand / RDMA file |
| Fibre Channel | - |
| Ethernet | 100 GbE |
| Host ports | 128 |
| Limits | |
| Volumes / LUNs | n/a |
| File systems | n/a |
| Snapshots | n/a |
| Management and support | |
| Management | HPE Data Services Cloud Console on GreenLake (Data Ops Manager, published APIs); HPE Alletra Storage MP Disconnected option |
| Power | 1,072 W |
| Positioned for | AI / ML, Analytics, Backup target |
| Warranty | 1-year parts-only (SSDs 1-year parts-only incl. wear-out); support contract mandatory |
| Design terms | Object store · LUN · Host group · Bucket replication |
| Data services | |
? Deduplication
✓ Compression
? Thin provisioning
? Snapshots
? Clones
? Asynchronous replication
? Synchronous replication
? Active-active / metro
✓ Encryption at rest
? Immutable snapshots
? Ransomware detection
? Quality of service
? Automated tiering
? Cloud tiering
? VMware vVols
? Kubernetes CSI
✓ Native file services
✓ Native object (S3)
? Non-disruptive upgrades
✓ Scale-out
✓ stated by the OEM · ✕ not offered · ? not stated | |
| Source | |
| Product page | https://support.hpe.com/hpesc/public/api/document/a50009215enw |
| Datasheet | https://www.hpe.com/psnow/doc/a50009215enw |