Hawthornehawthorne

// hawthorne — distributed object storage

Store data everywhere. Read it like it's local.

S3-compatible object storage on a global mesh of independent nodes. Objects are encrypted client-side, split into 80 pieces, and scattered across regions — so no single node ever holds a readable copy, and reads feel local wherever you are.

No egress fees. Roughly 80% under hyperscaler pricing.

quickstart →specifications
network.status● live
active_nodes
18412
objects_stored
1.41e9
bytes_stored
2.3 PB
median_read_ms
41
median_write_ms
63
pieces_per_object
80
rebuild_threshold
29
node_availability
0.9995
region_overlap
0
uptime_90d
99.983%

self-reported · telemetry sampled every 60s · values in SI/decimal

trusted_by[] =cloudwavevivintgabbtruenascaltechmeridian
[01]

products

two ways onto the mesh

object-storage

primary

S3-compatible storage spread across thousands of independent nodes. Drop-in for existing tooling — point your bucket config at Hawthorne and nothing else changes.

$ aws s3 cp render.mov \
    s3://media-archive/ \
    --endpoint https://gw.hawthorne.io

upload: ./render.mov  [80 pieces → 80 nodes]
done. durability 11 nines. egress $0.00

node-mount

fuse

Mount any bucket like a local disk. Open multi-gigabyte files without downloading them whole.

▸ MasterEdit_v3.mov4.1 GB
ColorGrade_FIN.mov2.8 GB
SoundMix_v2.aiff612 MB
[02]

how it works

encrypt · split · distribute · retrieve
  1. 01

    encrypt

    The object is encrypted on your machine before it ever touches the network. Keys stay with you.

  2. 02

    split

    Erasure coding breaks it into 80 pieces. Any 29 are enough to reconstruct the whole file.

  3. 03

    distribute

    Pieces are scattered across independent nodes in different regions. No node holds a readable copy.

  4. 04

    retrieve

    On read, the fastest 29 nodes answer in parallel. No syncing, no full download, no waiting.

OBJECT1.0 GB[ erasure: 80/29 ]24 nodes shown■ holds a piece□ standby
[03]

specifications

the fine print, up front
protocol
S3-compatible API
encryption
client-side, AES-256-GCM
erasure coding
80 pieces / 29 to rebuild
durability
99.999999999% (11 nines)
egress fees
none
regions
global — 18,400+ nodes
min node overlap
0 (no single node is readable)
SLA
99.95% availability
0%

lower cost

0%

less carbon

0k+

nodes

[04]

field reports

teams that stopped waiting
CASE-01
post-production studio

Editors pull 4K source straight from buckets. The render farm stopped waiting on syncs.

3.2× faster first-frame
CASE-02
SaaS platform

Cold archives moved off a single-region provider. Same durability math, a third of the bill.

−68% storage spend
CASE-03
genomics lab

Datasets shared across three continents. Mounting a bucket beat shipping drives entirely.

0 egress charges
bash — hawthorne

# spin up a bucket in under a minute

$ npm i -g hawthorne-cli

$ hawthorne init my-project

bucket ready · 25 GB free · no card required

create free account →talk to an engineer