Object RAID

PERFORMANCE, RELIABILITY, AND AVAILABILITY

Intelligent object RAID is at the heart of the PanFS operating system's ability to efficiently deliver high performance with uncompromising reliability and availability. While compute clients accessing Panasas storage simply see a single, high performance, local file system, PanFS is an object-based architecture where RAID levels are assigned on a per-file basis and all files are stored as objects. Objects combine the manageability advantages of traditional file storage with the high performance of block storage.

One reason why Panasas storage is so fast is that the PanFS object storage architecture eliminates the need for two common bottlenecks present in many competing platforms – the hardware RAID controller and the file server. With PanFS there is no longer a need for legacy hardware RAID controllers and RAID performance scales linearly with the number of compute clients. More importantly, reads and writes no longer have to go through a traditional file server because compute clients can access Panasas storage blades directly and in parallel, with performance scaling linearly with the number of storage blades deployed.

When it comes to reliability and availability, object RAID also has solid advantages when compared to pure file or block-based approaches. Instead of depending on legacy hardware RAID controllers that have no file level understanding, object RAID is performed on a per-file basis as an integral part of PanFS with three independent, complementary tiers of parity protection:

• Horizontal Parity maintains the reliability of the RAID group across multiple drives. Competing products based on hardware RAID controllers must reconstruct entire hard drives at a block level, blind to the files stored on them. With Panasas storage, data reconstruction occurs in parallel at an object level, restoring the system to a fault tolerant state at industry-leading speeds.
• Vertical Parity maintains the reliability of individual disk drives. It dramatically reduces the risk of dual drive failures by isolating and repairing media errors at the disk level before they impact the integrity of the horizontal parity.
• Network Parity maintains the integrity of the data path between the storage system and the clients. It automatically corrects any data corruption introduced by the network infrastructure by performing data integrity verification at the client node itself.

As a result, the PanFS storage operating system is the most comprehensive, scalable, and reliable high performance storage architecture available, making Panasas storage ideal for a wide variety of technical computing applications and big data workloads.