NetApp All-Flash FAS Arrays: A Deep Dive
IDC White Paper: How Flash boosts speed, performance and cost efficiency.View Now
Many vendors offer high-performance all-flash arrays (AFAs) to address high-throughput workloads with random data access patterns and low latency requirements, such as virtual desktop infrastructure and online transaction processing. AFAs offer performance advantages over hard disk drive-based systems for these workloads, yet many of them lack critical capabilities in one key area: enterprise data management.
NetApp offers all-flash FAS storage arrays that can be deployed as part of a unified scale-out architecture and that benefit from the industry-leading data management features of the Data ONTAP® operating system. The combination of flash technology and enterprise data management delivers advantages in three major areas: performance, availability and storage efficiency.
Scalable, Flash-Accelerated Performance for SAN and NAS
Increased Storage Performance
The latest generation of NetApp FAS systems leverages multiple CPU cores and software enhancements to unlock higher solid-state drive (SSD) speed than any prior FAS platform. Based on a PCIe, Gen3 architecture, the NetApp FAS8000 family can scale out to 24-nodes to provide millions of IOPS at submillisecond latency and support nearly 5PB of SSD capacity. In addition to new hardware options, other features enable NetApp FAS systems to optimize SSD performance, including:
- De-staged writes. By using the NetApp WAFL® (Write Anywhere File Layout) journaling system, a NetApp all-flash FAS eliminates write penalties associated with all SSDs by removing SSD writes from the critical latency path of the workload. It does so by writing first directly to system memory, a much faster process than writing to SSDs. All FAS systems use battery-backed, nonvolatile random-access memory (NVRAM) to protect incoming writes. When writes are received in memory from the host, they are logged into NVRAM. A write acknowledgement is immediately sent back to the host. This shows that the write has been completed. During a periodic consistency point, writes are then de-staged to the SSDs.
- Proximal data patterns. As writes are de-staged from memory to SSD, WAFL uses highly flexible SSD write allocation policies and a "proximal" write pattern to reduce the impact of random operations to the SSD. By using these practices, no blocks are permanently assigned to fixed disk locations. This approach optimizes the SSD’s write throughput and reduces the need for extra write cycles, which are an undesirable phenomenon associated with flash technology, often referred to as "write amplification." Reducing the number of writes and lessening the impact of random operations increases performance and extends the overall life span of the SSD.
- Read-ahead caching. A Data ONTAP custom read-ahead algorithm identifies “hot” data that is likely to be read most often. Before a read request comes from the host, the algorithm has already detected read patterns and pre-emptively staged to system memory those reads that are most likely to be accessed. Cached reads are then read directly from memory, with no need to access the data directly from SSD. This practice reduces the number of I/O requests that an SSD must serve directly, making more room to support heavy periods of SSD writes. Scheduling algorithms are also used to prioritize latency-sensitive reads over throughput-sensitive writes.
When deployed as part of a scale-out cluster, all-flash arrays benefit from the non-disruptive operations enabled by clustered Data ONTAP, which has been documented to provide greater than 99.999% uptime in production environments. The ability to move workloads within a cluster that can scale to 24-nodes enables several advantages. First, a wide range of operations can take place without the need for storage downtime, including performance balancing, hardware and software upgrades and major lifecycle operations. Second, data protection costs can be reduced by replicating data from all-flash nodes to hybrid nodes with a lower cost per terabyte. And third, workloads can be seamlessly moved between hybrid nodes and high-performance all-flash nodes to optimize price/performance in response to changing requirements.
NetApp SSDs are rated for high levels of daily write activity – warrantied and backed by a proven, worldwide support organization. It's also easy to monitor SSD status with a simple CLI command that provides useful information about SSD wear, such as:
- Percentage of device life already used.
- Percentage of spare data blocks already consumed.
- Threshold limit based on the percentage of spare blocks already consumed.
In the unlikely event an SSD wear threshold is exceeded, the Data ONTAP Event Management System sends a message and logs an AutoSupport event to enable preventative maintenance.
Flash Storage For Dummies, NetApp Special Edition: Discover how to optimize data performance and reduce the footprint of storage infrastructure in the data center.
Cost is among the primary concerns when considering all-flash arrays. Despite ongoing price reductions, SSD-based systems continue to command a significant premium over conventional HDD-based storage arrays and hybrid arrays that combine flash with HDDs. All-flash vendors are just beginning to introduce data reduction techniques such as deduplication and compression to offset the higher cost of SSDs.
However, the ability to deploy an all-flash array with Data ONTAP provides access to an even broader suite of storage efficiency technologies to reduce costs. NetApp all-flash FAS users can store useable data capacity that exceeds the raw storage capacity of the system by using the following efficiency technologies: RAID-DP®, Snapshot™ technology, Thin provisioning, Deduplication, Data compression, FlexClone® technology, Storage-efficient replication.
NetApp Flash Solutions
NetApp offers a broad range of flash-optimized storage solutions -- including hybrid flash arrays and all-flash arrays -- designed to increase application performance while maintaining high levels of reliability and availability.
For more information, visit our Tech OnTap® community for monthly updates on best practices, technical case studies, and in-depth interviews with engineering experts.
© 2014 NetApp, Inc. All rights reserved.
No portions of this document may be reproduced without prior written consent of NetApp, Inc. Specifications are subject to change without notice. NetApp, the NetApp logo and Go further, faster, are trademarks or registered trademarks of NetApp, Inc. in the United States and/or other countries. Click here for a full listing of NetApp trademarks. All other brands or products are trademarks or registered trademarks of their respective holders and should be treated as such.