luchschen_shutter - Fotolia

News Stay informed about the latest enterprise technology news and product updates.

All-flash NexGen arrays bring QoS to the forefront

The newly released NexGen arrays dynamically manage critical workloads with QoS policies. Spelling a move beyond hybrid storage, the all-flash arrays include SSDs, RAM and flash PCIe.

NexGen Storage Inc. today launched its first all-flash array, keeping the PCIe-based flash architecture and quality-of-service (QoS) capabilities from its hybrid arrays.

The new NexGen arrays use a combination of PCIe flash and solid-state drives (SSDs) for storage. The SSDs replace the hard disk drives used in its hybrid systems.

NexGen has two all-flash models, both including 2.6 TB of PCIe flash. The N5-1500 scales between 15 TB and 60 TB of raw SSD capacity, while the N5-3000 scales from 30 TB to 60 TB of SSD capacity.

Along with RAM, the PCIe flash and high-capacity SSDs give the NexGen arrays three tiers of flash. The systems contain 96 GB of RAM, and the PCIe flash is used both in cache and persistent storage. The SSD layer scales with 30 TB, 45 TB and 60 TB SSD capacity packs.

The NexGen arrays use the vendor's dynamic QoS with automated throttling to manage performance levels via preconfigured policies, adaptive queuing and prioritized active caching for data placement.

"It's an all-flash array and that isn't anything special. Everybody has all-flash arrays. What is important is the shift," said Chris McCall, vice president of marketing at NexGen. "We created software in an enterprise system that targets specific workloads, and we are moving beyond that now. It's about how to manage it more efficiently."

George Wagner, NexGen's senior product marketing manager, said the multi-tier, all-flash system supports ultra-low latency and high-performance capacity in a single array. The QoS allows performance-hungry workloads to get more predictable performance. The low latency reads and writes are handled by the PCIe flash, while the low latency read cache is done in RAM and SSDs handle high-performance capacity reads.

The NexGen arrays use the vendor's dynamic QoS with automated throttling to manage performance levels via preconfigured policies, adaptive queuing and prioritized active caching for data placement.

"Latency is still inherent in all-flash arrays," Wagner said. "Without QoS, there is no way to controls what workloads get impacted. The administrators need to know what kind of performance the applications need and preconfigure it. We can guarantee any volume assigned to mission-critical policies to ensure those [service-level agreements] will be met."

Mike Matchett, senior analyst at Taneja Group Inc., in Hopkinton, Mass., said the significant part of the all-flash NexGen arrays is how the vendor has brought its QoS to the new systems.

"Users want consistent, all-flash performance all the time," Matchett said. "But mixed workloads in flash will compete with all the flash resources. That is the challenge. [NexGen] has tweaked its algorithms to work across multiple tiers of flash. They have brought their intellectual property over to all-flash arrays."

NexGen began shipping hybrid arrays in 2011, touting its QoS and PCIe flash. Its PCIe flash provider Fusion-io acquired NexGen in 2013. After SanDisk acquired Fusion-io, it spun out NexGen in January 2015. NexGen is run by its original management team, including founders John Spiers and Kelly Long.

Next Steps

Features to look for in an all-flash array

Choose the right all-flash array for your needs

Is a hybrid or all-flash storage array best?

Dig Deeper on All-flash arrays

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

How much of a difference does the type of quality of service used in all-flash arrays make?
For Virtualized Workloads having a system that can provide QOS at the individual VM layer is a true differentiator and can only be delivered by Tintri.  Also, Tintri is a Flash first file system so your Virtualized Data is first stored on the highest performing storage level, insuring that the data the is written first will be read out of Flash.
I believe Tintri only supports virtualized workloads. Customers who aren’t 100% virtualized need their storage investment to accommodate more than just virtualized workloads. NexGen delivers QoS at the VM level (VVOLs), as well as to non-virtualized workloads. Additionally, NexGen employs TBs
of PCIe flash for all writes, which is much faster than SAS/SATA SSDs.

VVOLs hold a lot of promise but the technology still is in the development phase