Design & TechnologyA unique no-compromise design, efficient and scalable
Smart approach to object storage
Objects are stored within containers in a persistent three level massively distributed directory: there is no classic subdirectory tree, objects are stored in a flat structure.
The data query path is not dependent to the number of nodes and performance is not affected by the natural growth of the storage capacity.
A unique global namespace can store 1000s of Petabytes with linear performance.
Caches are everywhere and are readily available on the client side of the solution.
Massively distributed grid
Servers (> Requirements) are organized as a grid of nodes massively distributed.
Each node takes part in the directory and storage services.
Intensive metadata load is distributed, there is no Single Point Of Failure and it is resilient to node failures.
Unlike consistent hashing based systems there is no need to rebalance data: new servers can contribute immediately without any complex data movements.
A continuous process that:
- Collects the metrics from each node
- Computes a score in real-time for a particular node
- Distributes scores to every nodes and clients
On the fly best match making for each request.
Nodes are automatically discovered and used at their true capacity and for the task they suit best.
This allows human-free management, dynamic data placement and advanced load balancing for optimal perfomance.
Grid for Apps technology
A data computing framework is natively integrated inside the Grid.
It is aided by the Concience for maximum efficiency and performance.
Full-stack application back-ends can be built on the single scale-out storage platform itself.
It uses otherwise wasted resources and vastly simplifies load-balancing, multi-instance deployments and system orchestration.
Broad access layer
We natively integrate Amazon® S3 API, OpenStack® Swift API, HTTP REST/API and industry file sharing protocols NFS, SMB, FTP or local mountpoints through FUSE.
More than that, we bring storage closer to the applications. By using our native client SDKs, for Python, C, and Java, applications become part of the Grid, take part in data placement and achieve maximum performance thanks to point-to-point accesses to remote services.
Where does it run?Hardware
OpenIO is hardware agnostic, it runs on any x86 and ARMv7 server and natively supports mixed hardware within a single cluster.Software
OpenIO runs on Linux.
Packages ready for production:
- CentOS / RHEL 7
- Ubuntu 16.04
- Ubuntu 18.04