Enterprise object storage for your own data center: a full S3 API, erasure coding, and automatic healing, plus per-division quotas, usage metering, verifiable audit, offline licensing, and one console for every cluster.
7-day trial on 1 node, installed by our team in your data center
*Single node, development laptop, Docker. Methodology is stated under the benchmark chart.
Every object is split into 12 data shards and 4 parity shards across 4 nodes × 4 disks. When a node goes down, any 12 shards still rebuild the whole object.
Your client sends a PUT exactly as it would to AWS S3. No special SDK.
The object is cut per 1 MiB block into equally sized data shards.
Reed–Solomon computes parity. This setup is RS(16,12): 12 data, 4 parity.
Every shard lands on a different disk; no two shards share a disk.
Four shards vanish at once. Any twelve shards are enough to read the whole object.
When a replacement node or disk comes online, missing shards are recomputed and written back.
No extra operators or scripts. Everything ships and is supported together with the data plane.
Separate divisions, subsidiaries, or applications as tenants with their own capacity, bucket, user, and request-rate quotas. Usage is sampled every 5 minutes, aggregated daily and monthly, and can be reconciled at any time for reporting and internal chargeback.
Versioning, Object Lock, SSE-S3/KMS/C, presigned URLs, multipart, lifecycle, notifications, replication, STS.
Ed25519-signed license files for air-gapped data centers. When expired: 30-day grace, then read-only; reads are never locked.
IAM, buckets, lifecycle, replication, alerts, support bundles, and metering in one place, with RBAC and MFA.
Offline disks detected within seconds, replacement disks formatted and healed automatically, buckets stay readable when metadata quorum is lost.
Disk, quorum, capacity, license, and tenant quota alerts to webhooks, Slack, Teams, Telegram, or email.
Hash-chained daily audit logs; integrity is checked with a single command.
The benchmark harness, raw data, and client matrix live in the product repository, so you can rerun them on your hardware.
| Operation | Ratio | Inodix vs MinIO |
|---|---|---|
| PUT 1 MiB | 96% | 195 vs 203 MB/s |
| GET 1 MiB | 116% | 519 vs 448 MB/s |
| PUT 4 KiB | 186% | 2,782 vs 1,494 ops/s |
| GET 4 KiB | 126% | 6,655 vs 5,288 ops/s |
| LIST 1000 | 343% | 360 vs 105 ops/s |
Single node, single disk, Docker volume, neither system fsyncing per object, 32 workers × 15 seconds, Apple M1 Pro laptop. Not production hardware: the ratio is what matters. With per-shard fsync (Inodix's default), PUT 1 MiB is 79% of MinIO without fsync. Request the raw data
| Client | Make bucket | Small PUT | Multipart | GET + MD5 | HEAD | LIST | COPY | Presign | Versions | Tags | DELETE |
|---|---|---|---|---|---|---|---|---|---|---|---|
| AWS CLI 2.38.1 | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed |
| boto3 1.43.114 | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed |
| AWS SDK JS 3.115 | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed |
| AWS SDK Go v2 | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed | passed |
| rclone 1.75.2 | passed | passed | passed | passed | passed | passed | passed | passed | not supported by client | not supported by client | passed |
| s3cmd 2.4.0 | passed | passed | passed | passed | passed | passed | passed | passed | not supported by client | not supported by client | passed |
– = the client does not support that feature. Results from our client test matrix, run before each release.
ceph/s3-tests: 342 of 348 passing. The rest are Ceph RGW-specific and torrent features. See known differences from AWS
Storage nodes do not depend on the console. If the control plane goes down, your applications keep reading and writing.
A Go binary with no runtime dependencies: S3 API, Reed–Solomon erasure coding, Raft metadata on Pebble, healing, scrubber, lifecycle, replication, encryption, notifications.
Admin console with RBAC and MFA, tenants, metering, alerting, licensing, and encrypted support bundles; connects to many clusters over Admin gRPC.
There are no binaries to download. The Inodix team handles installation, the trial, and migration together with your infrastructure team.
Capacity, node count, networking, and security policy are mapped together with your infrastructure team.
We install the cluster on your hardware or Kubernetes, on site or remotely, including the console and alerting.
Your applications try the S3 API, tenants, and metering on your own data before you decide.
Data moves from MinIO, Ceph, AWS S3, or any other S3-compatible system: checksum-verified bulk copy, delta syncs, then a scheduled cut-over.
A license file is applied to the same cluster without reinstalling; support runs per contract.
Annual licenses include updates and support. Final rates are agreed according to scale.
7 days
at first install
For evaluating in your own environment.
For production
Per TiB + per node
annual license
For production in your data center.
Yes, for the APIs applications commonly use. It is tested with ceph/s3-tests (342 of 348 passing) and a matrix of six popular clients before release. Known differences from AWS, such as SelectObjectContent, are documented openly.
Our team plans and runs the migration with you from any S3-compatible system: an inventory of buckets and policies, a bulk copy with rclone or aws s3 sync (both in the tested client matrix) with checksum verification, repeated delta syncs while the old system still takes writes, then an endpoint cut-over in a window you choose.
No. Licenses are verified offline from a signed file. Activation through a license server is optional for internet-connected environments.
The cluster enters a 30-day grace period with warnings in the console. After that it becomes read-only: data is never deleted and reads are never locked.
No. Inodix is built from scratch (clean-room) and only uses permissively licensed dependencies: MIT, BSD, Apache-2.0, and MPL-2.0.
For production, at least 4 nodes with several data disks (xfs recommended) and NVMe for metadata. For a trial, one node with four disks is enough. Our team sizes the hardware with you during the assessment.
The Inodix team in Indonesia, in Indonesian or English. Encrypted support bundles from the console speed up diagnosis without opening access to your data.
Our team installs a 7-day trial on 1 node in your data center, with every S3 and console feature. The enterprise license is applied to the same cluster without reinstalling.
