DRACO 10 PB Object AI Storage

Overview

The DRACO 10 PB Object AI Storage is a scale-out, S3-compatible object storage platform for the AI data lifecycle — 10 PB usable (erasure-coded) of durable capacity for datasets, model registries, checkpoint archives and warm/cold tiers. It gives AI teams a private, exabyte-scalable object store behind their firewall, in INR, on a GST invoice, so training data and model artefacts live in-house with cloud-like S3 access.

Engineered as the capacity tier beneath the high-bandwidth NVMe parallel filesystem, it stores the bulk of AI data durably and cost-effectively with erasure coding, exposing the S3 API your pipelines and MLOps tools already use — delivered as one validated cluster with one warranty and one support contract.

Key highlights

• 10 PB usable (erasure-coded) · S3-compatible — exabyte-scalable durable capacity with the S3 API your tools already speak.
• Erasure-coded durability — high data durability without full replication overhead.
• Hybrid NVMe + HDD, scale-out — hybrid NVMe + HDD scale-out for cost-effective capacity with fast metadata.
• High-throughput Ethernet — high-throughput access for many concurrent clients.
• Capacity tier for AI — pairs with the NVMe parallel-FS hot tier; stage hot data up, archive cold data down.
• On-prem data sovereignty — datasets and model artefacts stay in-house; DPDP-friendly, air-gappable.
• Make-in-India OEM — predictable INR pricing, GST tax invoice (HSN 8471), pan-India onsite support, GeM-procurable.

Where it fits

• Dataset lake (primary): the durable, S3-accessible home for training datasets and raw data.
• Model registry & artefacts: versioned storage for model weights, checkpoints and experiment outputs.
• Warm/cold tier: the capacity tier beneath the NVMe hot tier; archive what isn’t being actively trained.
• Data sharing: S3 access for teams, pipelines and MLOps tooling across the org.

How it works

A scale-out object cluster stores data as S3 objects with erasure coding for durability, exposing the S3 API over High-throughput Ethernet. Capacity and durability scale by adding nodes. It pairs with the NVMe parallel-FS hot tier: stage active datasets up for high-bandwidth GPU feeding, keep the bulk durably here. Honest note: object storage is the capacity/durability tier, not the high-bandwidth GPU-feeding tier — pair it with NVMe parallel-FS for training I/O.

Industry use cases

• Government & national labs — sovereign data lakes and archives for national AI and HPC.
• AI/ML & foundation-model teams — durable dataset and model storage.
• BFSI & healthcare — data-resident storage under compliance.
• Media & design — large asset and archive storage.
• Energy & manufacturing — large datasets for AI/HPC.
• Neocloud / AI providers — the capacity storage tier for a GPU cloud.

Performance — and how to be sure

We don’t publish inflated peak numbers. The honest picture: 10 PB usable (erasure-coded) of durable S3 capacity. Want certainty? Request a free benchmark with your data and access pattern on this exact configuration before you buy; we’ll send back real throughput, durability and access latency.

Series & scale path

• DRACO (flagship storage tier) — this.
• Tiering: NVMe parallel-FS hot tier (GPU feeding) + object capacity tier (durable datasets/archives).
• When to step up: add nodes for more capacity and durability — talk to an architect about the tiering.

On-prem vs cloud — the TCO case

For AI data at scale, owning beats renting: cloud egress and per-GB costs add up fast against large datasets, and on-prem keeps data resident. RDP pricing is fixed in INR with a GST input-credit-eligible invoice — ask for a 3-year TCO comparison including egress savings.

Software & integration

Integrates with your stack: S3-compatible object (erasure-coded), standard S3 clients and MLOps registries, with monitoring and quota/multi-tenant management. Works with PyTorch/TensorFlow data loaders, Slurm/Kubernetes and common MLOps tooling.

Power, cooling & rack integration

A Scale-out cluster air-cooled system with redundant PSUs — specify rack power and cooling. (Exact power draw, BTU, airflow and footprint figures confirmed on the build sheet.) Full out-of-band management and drive hot-swap.

Deployment, warranty & support

• Made to order, built, racked and burned-in in India; lead time confirmed at quote.
• In the box: storage nodes, drives, rails, cabling, quick-start, and the configured object cluster.
• Onsite warranty + AMC with pan-India coverage, drive-replacement SLA and an RMA/escalation path (exact terms confirmed at quote).

Why RDP

14 years of Make-in-India infrastructure and 300,000+ devices shipped. Indian OEM, INR pricing, GST tax invoice (HSN 8471), pan-India onsite engineers, GeM availability, and DPDP / sovereign-AI-ready deployment.

Buy with confidence

This is an AI storage system, made to order — talk to an RDP solution architect, size capacity and tiering for your AI estate, get a 3-year TCO, and benchmark your own access pattern before you commit. Request a quote to begin.

Specifications