Tape is Back.
On-Premises S3 GlacierWithout the Lock-In.
On-premises S3 Glacier that writes directly to tape - no filesystem layer, no egress fees, no vendor lock-in. From terabytes to exabytes on any hardware you choose.
Built for the AI Era. Delivered on Tape.
An object-native archive - not an S3-to-file layer bolted onto a filesystem.
AI training sets, genomic data, imaging archives, and sensor logs are growing faster than any disk budget can absorb. XtreemStore gives those workloads a permanent home - scaling from terabytes to exabytes on LTO tape without breaking the S3 workflows that put the data there.
XtreemStore ingests, organizes, and writes data directly as objects - delivering S3 Glacier retrieval semantics inside your own datacenter, under your own control.
Why Now - The Return of Tape at Hyperscale
Tape is undergoing a structural resurgence. Modern rack-scale libraries - including high-density modular designs - are redefining what deep archive looks like. Capacity is growing exponentially, while power consumption and cost-per-terabyte continue to drop.
Meanwhile, the economics on the other side are breaking. Disk and server prices are climbing, supply remains tight, and hyperscale demand for GPUs and flash is pushing enterprise storage budgets past their limits. Cloud providers still charge significant egress fees, long-term retention costs are unpredictable, and organizations are actively repatriating cold data back on-prem.
This is creating a new architectural model: Fast storage stays online. Cold data moves to deep archive - where it remains retrievable, governable, and ready to work again when it is needed. Object workflows stay unchanged. No new translation layers.
XtreemStore was built specifically for this model.
The shift to object-native deep archive is no longer optional - it is becoming the default architecture for long-term data retention.
A Modern Approach to Deep Archive
Enterprise storage environments are no longer filesystem-centric. They are object-first, policy-driven, multi-tiered, and multi-vendor.
Most archive gateways store objects in arrival order. XtreemStore writes objects intentionally - based on metadata-defined relationships.
True On-Premises S3 Glacier
XtreemStore ingests, organizes, and writes data directly as objects - preserving the native structure of modern S3 workflows and delivering true Glacier semantics on-premises.
Why Object-Native Deep Archive
- Preserves S3 Glacier semantics end-to-end
- No protocol translation overhead
- Direct integration with lifecycle policies
- Simplified data path to tape
- Native S3 workflows remain unchanged
XtreemStore delivers true on-premises S3 Glacier - not a translation layer on top of other storage.
Why Gateway-Based Archives Add Complexity
- Additional protocol translation layers
- Extra metadata processing overhead
- More operational failure points
- Performance tied to gateway scaling
XtreemStore eliminates the gateway layer entirely by writing objects directly to archive media - preserving native S3 workflows without translation overhead.
Built to Work With Any Infrastructure
Any Tape Library
Supports all major LTO libraries, multi-library deployments, mixed-vendor environments. Tape pools can be dynamically assigned and expanded across libraries.
Any Disk Cache or Storage Platform
NVMe cache, SAN storage, NAS storage, distributed cache nodes. Functions as high-speed ingest buffer, usable near-line tier, and staging environment tuned for tape streaming.
Any Glacier-Compatible Workflow
Works directly with Ceph lifecycle policies, VAST Glacier integration, backup platforms, research pipelines, and custom S3 workflows using standard S3 Glacier semantics.
Scale Without Limits - To Exabytes
Linear scalability in both capacity and throughput - from terabytes to exabytes+ with no software-imposed limits.
Used in research archives, AI pipelines, backup retention systems, and long-term compliance storage environments.
Scale Up - Increase System Throughput
Add more tape drives, faster disk cache, or higher bandwidth networking.
Result: Higher sustained ingest and retrieval speeds.
Scale Out - Expand to Exabytes
Add additional nodes, libraries, and workers.
Result: Linear scalability. No architecture redesign required.
Deployed in environments managing hundreds of petabytes and scaling toward exabyte-class archives. Supports billions of objects across distributed archive infrastructures.
Typical deployments include multi-library environments supporting hundreds of petabytes of archived data across geographically distributed sites.
Designed to support billions of objects across distributed archive environments with linear scalability across drives, nodes, and libraries.
Policy-Driven Lifecycle Management
Intelligent Data Flow Pipeline
Policies define grouping rules, container size, movement timing, and storage placement. Tape operations follow configurable lifecycle rules.
Multi-Copy Protection and Geo-Replication
Up to Four Independent Tape Copies
Distribution across libraries or locations with independent durability zones for maximum redundancy.
Multi-Site Replication
Asynchronous replication to secondary XtreemStore clusters supporting distributed archive environments and disaster recovery workflows.
Metadata-Aware Data Organization
Objects can be grouped using project identifiers, dataset structure, metadata tags, or retention classification. This enables logical data segmentation, efficient tape utilization, and simplified compliance workflows.
High-Throughput Parallel Archiving
Objects are grouped into large optimized TAR containers for sustained streaming performance. Each LTO drive delivers sustained streaming performance in the hundreds of MB/s range, enabling aggregate throughput to scale linearly across multiple drives and nodes.
Hardware Freedom - No Vendor Lock-In
Choose any tape vendor, any library model, any disk infrastructure. Change hardware at any time without rewriting your archive. XtreemStore integrates - it does not replace your existing storage platforms.
Open Formats for Long-Term Durability
XtreemStore writes standard TAR containers on LTFS tape format. Your data remains readable without XtreemStore or any proprietary dependencies - protecting long-term data ownership.
Supports LTO-6 through current LTO generations, enabling long-term infrastructure compatibility and upgrade flexibility.
Designed for Layered Storage Architectures
Not all data is created equal. Hot working sets demand flash performance. Warm data needs accessible capacity. Cold and frozen data - which typically represents 70 to 90 percent of total volume - needs durable, low-cost retention without losing addressability.
XtreemStore is the Cold and Frozen tier - purpose-built to integrate beneath your existing Hot and Warm platforms without replacing them.
Access latency labels are typical ranges - actual times depend on infrastructure design, tape mount state, and object size.
Independent Tier Scaling
Scale Hot flash, Warm object storage, and Cold/Frozen archive independently. Add capacity where needed without over-provisioning adjacent tiers.
Cost Structure Matches Data Value
Pay premium rates only for data that needs premium performance. Move aging data down the stack automatically through lifecycle policies.
Protocol Consistency Across Tiers
S3 workflows remain unchanged as data moves between tiers. No re-architecting applications, no translation gateways, no broken namespaces.
Designed to Complement - Not Replace
XtreemStore sits beneath platforms like Ceph, VAST, Panzura, and Wasabi - extending their lifecycle policies into durable on-prem Glacier and deep archive tiers.
Built for Real Enterprise Workloads
Life Sciences
Genomics, proteomics, imaging
Medical Archives
PACS, DICOM with compliance requirements
Media & Entertainment
Large content libraries with metadata-driven access
Research & Government
Long-term retention and regulatory storage
Advanced Capabilities
Bucket-Based Archive Management
Native S3 buckets with bucket-level policy control
Tape Spanning
Seamless multi-tape support for large objects
Distributed Architecture
Modular services for scalable deployments
Air-Gapped Security
Offline tape isolation and ransomware protection
Operational Monitoring & Visibility
Real-time system visibility, job tracking, and tape activity monitoring provide operational transparency and simplified management
XtreemStore Architecture at a Glance
- Pure object-native deep archive
- True on-premises S3 Glacier semantics
- No S3-to-file translation layer
- Supports any tape library vendor
- Up to 4 independent tape copies
- Multi-site replication support
- Open TAR / LTFS data format
- Linear scalability to exabytes+
Comparison Snapshot - Modern Archive Architecture
| Capability | XtreemStore | Bundled Appliances | Filesystem Gateways |
|---|---|---|---|
| Pure Object-Native Archive | ✔ | Limited | No |
| Works With Any Tape Library | ✔ | No | Often limited |
| Works With Any Cache Storage | ✔ | Restricted | Varies |
| Up to 4 Tape Copies + Geo-Replication | ✔ | Limited | Usually not native |
| Open TAR / LTFS Formats | ✔ | Often proprietary | No |
| No Filesystem Namespace Limits | ✔ | N/A | Creates limits |
| Scales Linearly to Exabytes | ✔ | High cost scaling | Challenges at scale |
| Requires S3-to-File Translation Layer | NO | Often | Yes |
| Hardware / Vendor Lock-In | NONE | High | Medium-High |
Frequently Asked Questions
Is my data readable without XtreemStore?
Yes. Data is written in open TAR and LTFS formats, ensuring long-term readability independent of proprietary software.
Can I mix tape vendors and libraries?
Yes. XtreemStore supports multi-vendor and multi-library environments with dynamic pool management.
How does XtreemStore compare to cloud Glacier?
It provides equivalent S3 Glacier semantics and lifecycle policies - without egress fees, unpredictable retrieval costs, or loss of infrastructure control.
How large can XtreemStore scale?
From terabytes to exabytes and beyond. Capacity growth is driven by your infrastructure - not software limits. It handles billions of objects with linear scaling.
Does XtreemStore require proprietary connectors?
No. XtreemStore integrates using standard S3 Glacier-compatible APIs and lifecycle policies. No proprietary translation layers or custom connectors required.
Can I integrate with existing Ceph or VAST environments?
Yes. XtreemStore integrates directly into existing lifecycle workflows without requiring architecture redesign. Your applications continue using standard S3 Glacier API calls.
Does XtreemStore require a filesystem layer?
No. XtreemStore operates as a pure object-native archive platform and does not depend on filesystem namespaces or inode structures.
Ready to See XtreemStore in Action?
See how your environment can scale to exabytes while preserving modern S3 workflows and maintaining complete control.
Learn how metadata-driven grouping makes deep archive data usable - even from tape.
This technical guide explains how metadata-driven grouping enables deterministic archive layout, faster retrieval, and scalable deep archive design.