NVMe all-flash storage is providing IT decision-makers with an unprecedented opportunity to achieve major breakthroughs in performance, simplicity and consolidation. Unlike earlier all-flash arrays, which used older interfaces and protocols designed for spinning disks, NVMe arrays use a host controller interface and protocol designed specifically to accelerate the performance of all-flash arrays.
As IDG notes, NVMe “is shaking up the enterprise storage industry.”1 One study predicts that 40% of all-flash arrays will be NVMe by 2020.2 According to Gartner, NVMe and NVMe over Fabrics (NVMe-oF) represent a new class of storage, called “shared accelerated storage.”
“NVMe-oF-based solutions minimize the protocol overheads and offer dramatically higher queuing capacities, resulting in lower latency profiles and the elimination of I/O bottlenecks, while optimizing data center footprints,” it says.3
When any new technology comes along with high expectations, decision-makers have questions about when it is appropriate to deploy it, how much it will cost, how it will impact existing operations and whether the high expectations are real or just marketing hype.
Although NVMe-equipped products have been on the market since 2015, the reality is that the technology is just now making its way into the consciousness of many IT decision-makers. Here, we address some of the most common and critical questions IT leaders and admins are asking.
Is the buzz around NVMe real or hype?
Simple answer: It’s real. NVMe arrays are up to two times faster than previous-generation all-flash arrays, with latency as low as 250 microseconds. They are available with ultra-dense modules that provide up to 3 petabytes of effective storage in a 6U rack. These arrays represent the foundation for NVMe-oF technology, which is uniting storage area network and direct-attached storage into a single storage architecture, delivering the simplicity and performance of DAS with the capacity, efficiency and reliability of shared storage.
Isn’t NVMe much more expensive than existing SAS all-flash arrays?
It has been more expensive, and most vendors still charge a premium for NVMe, offering it as a cache or high-performance tier. This helps them slow the pace of the market so it doesn’t affect their installed bases. Pure Storage® is the lone vendor that has taken a different approach. Pure has been using NVMe in its arrays since 2015, and 50% of its revenue comes from products shipping with NVMe technology. Pure does not charge a premium for NVMe. Customers can purchase an enterprise-class all-flash array with either NVMe or non-NVMe technology for the same price, using the same architectural model and Purity software.
What are the characteristics that make NVMe enterprise-ready?"
“Enterprise-ready” means different things to different people, but what it comes down to is this: Are you fully comfortable using the technology for your most business-critical enterprise applications? With NVMe from Pure Storage FlashArray™ storage arrays, you can be assured that enterprise-readiness is built in. In fact, NVMe arrays can deliver improvements not just in the performance of key applications, but also in reliability, availability and business continuity. FlashArray//X arrays from Pure Storage deliver ultra-dense consolidation with proven 99.9999% availability.
If our organization moves to NVMe, does that mean we will need a whole new storage architecture?
Again, it depends on the vendor. Pure Storage engineered its current line of NVMe-enabled FlashArrays using the same architecture and Purity software as its non-NVMe arrays. This means consistency across the entire product line, enabling you to upgrade with NVMe non-disruptively for applications that can benefit from much greater performance and density, while keeping your existing arrays for less demanding applications. By using the same modular, stateless architecture, NVMe arrays take advantage of some of the key features and functions that are consistent across the Pure product line, namely:
- Simplicity: NVMe arrays are no more complex to manage, install, upgrade and scale than non-NVMe arrays from Pure Storage. This means you can be up and running in less than a half hour, with just six cables, API-based automation and no tiering or performance-tuning requirements.
- Consolidation: The unprecedented performance density of NVMe enables you to consolidate mixed workloads in your hybrid or private cloud with no fear of I/O contention. Additionally, you can take advantage of Pure Storage’s industry-leading data reduction, which is typically two times better than competitive solutions, thereby lowering your capacity requirements.
- Evergreen Storage: FlashArray//X arrays with NVMe are available with Pure’s Evergreen™ Storage, which means you can purchase them using an Opex model—enabling you to expand and improve performance, capacity, density and features without downtime, forklift upgrades, performance impact or data migrations.4
NVMe is a technology of the future that is available in the present. Implemented properly and strategically, it can provide your organization with competitive business advantage through higher performance, consolidation, cost savings, speed to market, greater reliability and increased availability. What are you waiting for? Now is the time to get smart about NVMe and what it can do for your organization. You can start by visiting Pure Storage.
1 “What Is NVMe, and How Is It Changing Enterprise Storage,” Network World, June 20, 2018
2 “G2M Research NVMe Ecosystem Market Sizing Report,” G2M Research, September 2016
3 “The Future of Hyperconverged and Integrated Systems Will Be Shaped by Shared Accelerated Storage,” Gartner, May 25, 2017
4 “Evergreen Storage Service (ES2),” Pure Storage