What is SGI OmniStor?
SGI OmniStor is an industry-leading object storage solution to provide disk-based storage with cloud-like scalability but also enabling the flexibility and performance that is normally associated with conventional RAID platforms.
Unlike many object storage solutions with performance only suitable to slow archives, SGI OmniStor enables the best of both worlds: disk-based object storage based upon commodity hardware to enable public and private cloud architectures, while also enabling high performance NFS/CIFS and block storage access. And native support for SGI DMF tiered storage virtualization, SGI OmniStor enables IT managers the cost and data protection advantages of scale-out object storage architecture while also bridging into conventional file system infrastructures. And this is possible at a lower cost than public cloud providers like Amazon S3.
Redefining Software-Defined Storage
SGI OmniStor is a software-defined storage solution that is deployed on a several choices of commodity hardware to delivers scalability, flexibility, performance and cost efficiency. SGI OmniStor can be built on virtually any x86 servers, transforming a server farm into a scale-out storage pool that can be integrated with enterprise and cloud applications to support a range of workflows.
Applications can be connected to SGI OmniStor using any of a variety of interfaces including HTTP/REST, the Amazon S3 API, a dedicated OmniStor API, Openstack Swift, and the SNIA CDMI standard. SGI OmniStor can also provide NFS and CIFS/SMB access for conventional NAS connectivity, enabling the solution to be used as scale-out file system for on-premise usage.
SGI OmniStor's state-of-the-art software provides many key features for new cloud-enabled data centers. It uses a peer-to-peer model and establishes an end-to-end parallelism between the clients systems and storage nodes in the infrastructure. Data is protected via replication, both local or geographically distributed, and also with proprietary erasure coding technology called ARC. ARC tolerates multiple simultaneous failures without losing data or access to data. The default configuration for ARC offers support for up to four failures with a maximum additional hardware overhead requirement of 30%.
SGI OmniStor storage is a Distributed System made of independent nodes that can be replaced over time as new technology becomes available.
- Object based
- Distributed system
- Distributed metadata
- No single point of failure
- No limit to expansion
- Performance comparable to conventional block storage
- Lower TCO than public cloud providers
Configurations for Capacity and Performance
SGI OmniStor can be deployed in multiple different configurations, depending on the size of the environment, the growth rate of data, but also the performance level required.
|File System Configuration
The File System configuration is geared for the higher performance environments, including those with NAS workflows on industry-standard protocols. The File System configuration is a mix of SSDs for fast metadata operations combined with SAS drives for reliability.
As with all OmniStor configurations, after an initial 6-node Ring is established, additional nodes can be added at any time without limitation. The system will automatically incorporate the new nodes and capacity into the Ring, and load-balance all of the data across all nodes. This is true whether it is 500TB, 100 petabytes or more, in a single location or distributed globally. There is no limitation to the scale of this system.
The Volume configuration is geared for the optimizing the cost of large volumes of data which might have lower performance requirements. The Volume configuration eliminates SSDs, adding additional HDD capacity to each node.
As with the other configuration one HTTP/REST connector runs on every server.
SGI OmniStor is the perfect architecture to address high demand cloud computing and storage challenges. This integrated solution offers the most scalable disk-based storage solution on the market today.