Exceptional Performance = Greater Business Benefits
Storage becomes the weak link in the overall performance chain when client machines wait on storage systems to input and output data. Exceptional storage performance is necessary to accelerate workflows for greater discovery, innovation, productivity, and profitability. Achieving exceptional storage performance at the scale required in technical research and enterprise environments can be a challenge. Legacy NAS systems can often scale capacity, but do not always deliver performance at scale, impacting applications as client nodes wait on storage for read/write operations. When workflows take longer to complete, productivity lowers in lock-step.
Greater business benefit is derived when workflows complete faster. Higher performance yields faster time-to-results, faster discovery and analysis, more innovation, and deeper business insight.
ActiveStor® Delivers Extreme Performance at Scale
The ActiveStor® hybrid scale-out NAS appliance eliminates the scalability and performance bottlenecks found in legacy and some other scale-out NAS systems. Through an innovative, next-generation architecture, ActiveStor accelerates time-to-results and drives exceptional aggregate mixed-workload performance at scale. This results from:
- Parallel data transfer
- Fully parallel file system with parallel data path
- Exceptional data transfer speed vs. legacy and serialized architectures
- Flash Acceleration
- High IOPs for small file workloads
- Increased metadata performance for overall faster file processing
- Direct data access
- Most file processing (metadata) operations are not in the read/write data path – resulting in exceptionally fast file data transfers
- Automated load balancing
- Optimized performance and elimination of hotspots
- Linear performance scaling
- Avoids the need for multiple islands of storage
- Elimination of file processing bottlenecks
- Parallel data paths
- Fast storage components
- Performance balancing
- Linearly scalable architecture
- Flash acceleration
Parallel Architecture with Direct Data Access
Similar to how a multi-lane freeway enables greater traffic flow, the PanFS® storage operating system uses parallel data paths to substantially increase data throughput to and from storage. DirectFlow® further enhances performance by processing file metadata outside the data path instead of in-band as is seen in many competing scale-out architectures.
PanFS also eliminates the performance scalability limitations of hardware RAID. Each client node calculates per-file RAID, so RAID performance scales with the number of clients accessing storage. RAID performance is driven across an entire compute cluster — avoiding bottlenecks by NAS heads, NAS clusters, and hardware RAID controllers.
Performance scales seamlessly. Simply add more shelves to non-disruptively and linearly scale out performance. Shelves can be configured to accelerate small file I/O, large file streaming, or a mix of both file workloads. Linear performance scale-out is accompanied with automatic load balancing to optimize performance and eliminate performance hotspots.
MAXIMIZING SCALE-OUT NAS PERFORMANCE
Several factors drive scale-out NAS performance:
These factors are known by all scalable NAS vendors, but integrating all of these factors for maximum aggregate performance requires a next generation scale-out NAS architecture that effectively leverages all performance factors.