Encrypted backup for datasets of any size.
Back up folders, build disk images, and host AI training datasets on Usenet — then stream-write them straight to a target disk or model server on restore. Every archive opens as a browsable network share you can mount, extract, or migrate without staging the full dataset locally first.
Tired of per-GB cloud egress and unpredictable storage bills on your backups and datasets? Switch to Usenet Share.
usenet://8fb92a@alt.binaries.data/71029#k=4f82a1c9e823b10
What it's for
Backup, storage, archives, migration.
Encrypted backup of a 400 GB project
No size cap, no per-GB cost. Restore the archive on any machine you own — mount it as a network share, pull individual files, or restore the whole set.
Create, mount & deploy disk images
Build raw / .img / .iso disk images, host them on Usenet, then mount, extract, or stream-write them directly to a target disk on restore. No local staging required.
Store & stream training datasets
Keep multi-terabyte AI training datasets in encrypted remote storage and stream them straight to a model server or training node — never stage the whole set locally.
Long-term encrypted archive management
Data is encrypted on your machine before it leaves. The app tracks provider retention and warns you before anything nears expiry so you can re-publish.
How it works
Six steps from source to restore.
Pick source
Folder, disk image, or AI training dataset.
Encrypt + pack
AES-256-GCM. Big files chunked, small files packed.
NNTP upload
60 concurrent connections to your Usenet provider.
Archive URI
One tiny link points to the encrypted archive + key.
Mount archive
Open the archive like a browsable network share.
Restore / stream
Extract, verify, or stream-write to disk / model server.
* You bring your own Usenet provider (Newshosting, Tweaknews, Eweka, …). See Providers.
The app
Every surface — real screenshots.
Backup & upload pipeline
Index → segment → encrypt → upload → verify → commit. Live throughput, transferred bytes, and a real folder tree of what's going into the archive.
Archive browser (network-share view)
Open an archive and browse it like a mounted network share. Pick a file, a subfolder, or the whole archive. Parallel restore with per-file SHA-256 verify + auto-repair.
Stream disk images straight to disk
Stream a disk-image archive range-by-range straight to a target block device on restore. No local staging, checkpointed, byte-verified on completion.
S3-compatible remote storage
Point rclone, aws-cli, boto3 or s3cmd at your local gateway to back up or migrate data to Usenet. Buckets map to archives, keys map to files. Zero per-GB egress.
Verified at scale
Numbers from the latest backup + restore run.
Backed up & restored end-to-end
Latest run: a 20 TB dataset backed up and restored clean — large streaming files, packed small files, unicode names, symlinks, hardlinks, boundary sizes.
Files in a single archive
Smart small-file packing bundles millions of tiny files into single Usenet articles, then reconstructs them perfectly on restore.
Articles stored, zero corruption
Every file SHA-256 hashed pre-upload and re-hashed on restore. Missing articles auto-repair from redundant copies.
Concurrent NNTP connections
Custom-modified NYUU upload + native C++ restore engine. Matches per-IP caps at Newshosting, Tweaknews, and Eweka.
Capabilities
Built like infrastructure.
Browse an archive like a network share
Open an archive and you see a real folder tree — directories, filenames, sizes. Restore one file, a subfolder, or the whole archive. No NZB, no manifest.
Disk images of any size
Create .iso, .img, or raw disk images and store them on Usenet. Mount, extract, or stream-write them straight to a target disk without local staging.
Write to Disk
Pipe an archive or disk image directly to a target disk or block device on restore. Zero local staging. Pause/resume + crash-resume built in.
Stream training data to model servers
Keep multi-terabyte datasets in encrypted remote storage and stream them to a training node or model server as a mounted archive. No local copy, no cloud egress.
Smart small-file packing
Millions of tiny files would mean millions of tiny uploads. Usenet Share packs files under ~500 KB into single articles, then unpacks them perfectly on restore.
End-to-end encrypted
AES-256-GCM. Data is encrypted on your machine before it hits your Usenet provider. Without the archive's key, the bytes on Usenet are random noise.
SHA-256 verified at every step
Every file is hashed pre-upload and re-hashed on restore. Mismatches trigger automatic re-fetch from redundant copies.
AI Dataset Streaming
Train on multi-terabyte datasets without staging a single byte.
Store raw datasets, model weights, and checkpoints as encrypted disk images on Usenet. Mount them on a training node and stream samples straight into your data loader — no local copy, no cloud egress, no per-GB retrieval fees.
Pack your dataset into a disk image (.img/.iso) or a folder archive. Small files auto-pack; large shards stay contiguous for sequential reads.
AES-256-GCM on your machine, then uploaded to Usenet with SHA-256 integrity and redundant article copies.
Point your GPU box at the archive URI. It appears as a browsable network share or a mounted block device.
Your data loader reads samples on demand. Prefetch + read-ahead cache keeps throughput saturated across epochs.
- Line-rate
- Saturates a 1–10 Gbps NIC on a well-provisioned Usenet provider.
- <100 ms
- First-byte latency to a mounted sample with read-ahead warmed.
- Zero egress
- No per-GB retrieval fees. Your provider bill is a flat monthly rate.
- Any size
- Single archives to 100+ TB. Stream a shard, an epoch, or the entire set.
- Train LLMs on raw text corpora
Mount a 20 TB tokenized corpus and stream shards into your DataLoader. Swap corpora by pointing at a different archive URI.
- Fine-tune vision models on image archives
Ship a disk image full of WebDataset .tar shards. The training node reads them sequentially without unpacking to local disk.
- Restore model weights & checkpoints
Stream-write a checkpoint image straight to a GPU node's local NVMe when spinning up a new training run.
- Distribute datasets to multiple nodes
One archive URI, N mount points. Each training node pulls what it needs — no central data server to run out of bandwidth.
Throughput and latency depend on your Usenet provider, network path, and archive layout. Numbers above reflect typical results on a 10 Gbps link with a tier-1 provider.
vs. the alternatives
Backup & storage stack comparison.
| Capability | Usenet Share | Backblaze B2 | AWS S3 Glacier | Dropbox Business | Self-hosted NAS | rclone + S3 |
|---|---|---|---|---|---|---|
| Max archive size | Unlimited | Plan-dependent | Unlimited | Plan-dependent | Disk-limited | Unlimited |
| Disk-image storage & stream-restore | Yes (stream-write to disk) | No | Manual | No | Manual | No |
| AI/ML dataset streaming | Yes (stream to model server) | No | No | No | Partial | No |
| End-to-end encrypted (client-side) | Yes (AES-256-GCM) | Optional | No (server-side) | No (server-side) | Depends on setup | Optional |
| Per-GB egress / restore cost | $0 (flat Usenet fee) | Included / per-GB | $0.09+/GB + retrieval | Included | $0 (LAN) | ~$0.09/GB (AWS) |
| Speed on big datasets | Fast (60 conns) | Fast, paid | Hours to restore | Moderate | LAN-fast | Fast, paid |
| Archive access model | Mount as network share | Web / client | Job-based restore | Web / client | SMB / NFS | CLI-only |
| Browse + selective restore | Yes (folder view) | Yes | Manual | Yes | Yes | Yes |
| Data durability | Provider retention + warnings | Provider SLA | Provider SLA | Provider SLA | Your hardware | Provider SLA |
| Self-hosted control | Yes | No | No | No | Yes | Partial |
| Resume after crash | Yes | Partial | Manual | Yes | Yes | Manual |
| Privacy from storage vendor | Total | Depends on key mgmt | None | None | Total (on-prem) | Total (client-encrypt) |
Pricing
Buy once. Own it.
Full app + updates for a year. Ideal to try a full project cycle.
- ▸All features included
- ▸1 activated device
- ▸12 months of updates
- ▸Paddle checkout
- ▸License emailed instantly
Same full app, two full years of updates. ~$2.08/month.
- ▸All features included
- ▸1 activated device
- ▸24 months of updates
- ▸Auto-renews via Revolut
- ▸Priority email support
When these 9 are gone, Lifetime closes forever.
Pay once, own the app. Every update, for the life of the product. Strictly limited to 100 customers — once they're gone, they're gone.
- ▸All features included
- ▸1 activated device
- ▸Lifetime updates
- ▸Self-serve device move
- ▸Priority email support
Back up your terabytes.
Windows, macOS, Linux. License from $29.99. 1 device per license.