Introducing Cach Efx Jan 08[Go]
Download as PPT, PDF•0 likes•300 views
The document introduces CACHEfx scalable caching appliances which use high-speed memory to overcome disk-based I/O bottlenecks and accelerate data center systems. CACHEfx appliances seamlessly integrate into existing infrastructures, initially focusing on accelerating NAS/NFS and clustered file systems. They deliver high performance in terms of low latency, high IOPS, and high throughput while supporting thousands of clients and scaling to multiple terabytes. Production customers have seen up to 5x application performance increases and 85% reductions in filer load with CACHEfx appliances.
Introducing Cach Efx Jan 08[Go]
- 1. Introducing CACHEfx Scalable Caching Appliances January 2008
- 2. Current Trends Creating I/O Bottlenecks Current trends driving painful storage problems Storage bottleneck trends Shared I/O applications File-content explosion Server virtualization Web-scale applications Oracle on NFS/DirectNFS Traditional solutions are no longer effective Disk over provisioning Server or subsystem caching
- 3. A New Approach: Centralized Storage Caching Dramatically increase performance IOPS, low latency, throughput Reduce over-provisioning Less power, space, cooling Guarantee service levels Install seamlessly Maintain storage processes (snapshots, mirroring, archive) Scale as a shared network resource Cached data served 10-50x faster from memory Clustered File System Gear6 Scalable Caching Appliance
- 4. Introducing CACHEfx Scalable Caching Appliances Use high-speed memory to overcome disk-based IOPS and latency limitations Accelerate data center systems and applications by seamlessly eliminating I/O bottlenecks Fit existing infrastructure Initial focus on NAS/NFS Support clustered file systems Deliver true performance Low latency High IOPS and high throughput Support thousands of clients Scale to multiple terabytes Close the performance gap Scalable Caching Appliances
- 5. Delivering Results for Production Customers! 3x Increase in project capacity with no additional storage hardware 5x Application performance increase 85% Reduction in filer load (IOPS and throughput) 100% Client utilization achieved, up from 60%
- 6. Positive Impact of Centralized Storage Caching Virtualization Databases File-centric data stores Shared I/O Applications Electronic design Animation Financial analytics Energy Media delivery Biotechnology New virtual machines Concurrent clients Random access Transaction intensive Large data sets Peak loads with SLAs Performance and acceleration Reduce over-provisioning Quality of Service Investment protection Broad Market Applicability Pervasive Data Center Trends Consistent Benefits
- 7. CACHEfx Appliance and Reflex OS Reflex OS™ Coherent memory aggregation Parallel I/O and real time response for scalable I/O bandwidth Intelligent cache services Appliance Performance From 250,000 to millions of IOPS Microsecond response time Scales linearly Network From 1.6 Gigabytes per second 10 Gigabit Ethernet connectivity Capacity Starting at ¼ TB. More capacity via real time compression. Scale to 5TB Supported Protocols NFS and clustered file systems Other protocols planned Virtualization Performance Cache Services Reflex OS
- 8. Deployment Use Cases Problems I/O constraints Poor utilization Varied service levels Solutions Application acceleration Efficient performance Guaranteed storage quality of service Benefits Shorten processing times Simplify configurations Easily manage peak loads No infrastructure change Supercharge all NAS/NFS and clustered file systems CACHEfx™ NFS Servers/Storage (clustered file systems) NFS Clients Application and Database Servers Ethernet/IP Network Direct data path available anytime Performance and Acceleration Reduce Over-provisioning Quality of Service
- 9. What Customers Are Saying About Gear6 : Gear6 Confidential Accelerating storage for real time application performance
- 10. Backup Slides
- 11. Caching and Namespace Performance Low Latency IOPS Throughput Global Namespace Memory Disks Shared Namespace File System Performance Capacity Caching performance Namespace performance
- 12. Current Ineffective Performance Approaches Latency and IOPS limited by disk New hardware, file system, storage software Great for capacity, not performance Scarce memory resources Limited capacity Difficult to scale and share Architecturally complex Latency and IOPS limited by disk Over-provision for performance Frequent “Hot Spots” High CAPEX and OPEX Parallel / Clustered File Systems Device Caching Traditional NAS