IT Focus Area: infrastructure operations
May 18, 2015
Making the Business Case for Flash Storage in Your Environment
In the past, the use cases for the highest-performance technologies have been limited to niche and uncommon situations where their high costs could be justified by unusually large returns.
Flash was one of these storage technologies.
However, over the past decade, prices for flash storage have decreased from more than $1,000 per gigabyte (GB) to in some cases less than $10 per GB. This decrease in price and the coming mainstream application of deduplication and compression—uniquely suited for flash-based storage—are going to further reduce the cost of this technology and bring extreme levels of performance to the masses.
This has not gone unnoticed in the industry.
Performance, Not Capacity, Drives Storage Costs
In today’s IT environments, capacity is not the primary driver for storage costs. If it were, enterprises would be quite comfortable with the sub-$1/GB costs of low-cost, low-performance near-line capacity available today. Capacity means little if it can’t support the speed required by the business. As a result, it is performance that drives the cost of primary storage today.
Over the past decade, compute performance has increased more than 8x and networks are 100x faster. While density of mechanical storage has increased 8x over the last 10 years, the performance characteristics have largely stagnated due to mechanical limitations. To overcome this challenge, storage manufacturers and enterprises have taken advantage of flash and other technologies to address the need for increasing performance requirements, while still reducing enterprise storage costs. This has been made possible by pooling and automated tiering intelligence, enabling storage systems to dynamically concentrate performance-hungry application data segments on flash storage while placing capacity-oriented, infrequently accessed data segments on mechanical storage.
Band-Aid vs. Cure for Reducing Storage Costs
These methods are better than nothing, but you are still limited by the size of the storage array's cache and the physics of mechanical spinning disk that can result in inconsistent performance depending on where the data is stored within the array. This is just a band-aid approach to the core problem. Sometimes spinning disk just isn't fast enough. Today, new breeds of “superefficient” flash systems combine high-performance flash and cutting-edge data-reduction technologies to bring unprecedented opportunities for reducing cost and a cure to the problem. Many companies have already begun to replace multi-million-dollar traditional storage solutions with flash storage systems that provide orders of magnitude better performance at less than 25 percent of the cost.
The cost benefits of flash storage go beyond the declining acquisition costs. Due to flash storage's unique physical attributes, deploying flash in the typical enterprise can result in substantial power, cooling, and floor space reduction in the data center. Additionally, the decrease in latency can increase the central processing unit (CPU) utilization of connected systems. Depending upon the software running these applications, this can amount to material savings from software licensing and related maintenance costs.
Performance or Capacity Constrained?
The reality of whether flash storage is right for your environment is based upon your use case—whether you are performance or capacity constrained. Flash storage is still relatively expensive for the capacity you get, so the ideal use case is for applications where performance is the driving factor. Consider the following illustrations.
Relative Costs for Performance-Constrained Environments
As you can see above, in use cases where performance is critical—for example, virtual desktop infrastructure (VDI), enterprise resource planning (ERP), e-commerce, or data analytics systems—flash-based storage has a much more favorable total cost of ownership (TCO) pattern as the performance demands increase. This is due to the fact that for traditional spinning disk, more and more drives must be deployed to service the required input-output (IO). This results in unusable capacity, wasted data center floor space and power, and a much higher TCO than required.
Relative Costs for Capacity-Constrained Environments
The second illustration shows the opposite effect. In use cases where performance is not a critical factor—for example, long-term object or file archives, replicated disaster recovery storage or backup storage—flash-based storage actually costs more. This is due to the much greater densities available in the traditional spinning disk form factors at a much lower price point per terabyte (TB). Deploying flash in these types of use cases results in wasted performance and a higher TCO than necessary.
See Where Flash Storage Makes Sense
Flash-based storage may seem expensive, but for those workloads that require high performance, the overall total cost of ownership can be attractive. Additionally, pricing for these systems dropped to a level where flash can be easily justified for the right use case. Performing a TCO analysis of your application environment is an easy way to gain visibility into the cost of storage performance and potential savings opportunities through flash storage. Our experience modeling flash TCO justifications has demonstrated significant multi-year savings opportunities for many enterprises. Now is the time to start considering where flash storage can yield savings in your own business as the cost of this technology continues to decline.