What’s Holding Back the All-flash Data Center?

Flash has certainly become more affordable over the past several years and we continue to see more and more storage products introduced that include flash in one form or another. But most flash implementations are hybrid in nature, requiring that the ‘right’ data be positioned in flash at the right time. This usually means caching or tiering algorithms to move that data around or simply pinning an application’s entire data set in flash while it’s being run.

There is another alternative to this complexity of worrying about which data is hot and which is cold and whether there’s enough flash capacity available at the right time; the All-Flash Array. These arrays greatly simplify this entire data management process since everything that’s stored on them is accelerated. They’re pretty much a ‘set and forget’ solution that eliminates performance problems, in most cases.

Is it really that simple?

While flash has come down in price, generally, the concept that flash is expensive still pervades, and so an all-flash system must be even more expensive. This perception was evident in a recent Storage Swiss webinar where we conducted a poll asking viewers “What’s holding you back from all-flash?” – 75% said it was cost.

But is this concern about high cost accurate, or do IT managers need to get past the old notion of raw acquisition $/GB? And what about reliability?


It’s time to look at total costs for the system, which means ‘effective’ capacity costs and especially, operational costs. Flash consumes roughly 5% as much energy as the same capacity of disk space and as little as 10% of the rack space in the data center. These are costs that are ongoing – month in and month out. Another operational cost is administrative overhead.

Early on flash provided management relief from legacy disk array performance techniques like wide striping and short stroking. The amount of effort required to keep these systems tuned was significant. Compared with hybrid flash and disk arrays, all-flash also requires much less administration, since there’s no caching or tiering needed to keep the right data on flash. There are also no cache misses.

In addition, all-flash arrays leverage technologies like deduplication, thin provisioning, snapshots and clones much better than disk systems do. This improves the data efficiency of the system and creates a larger ‘effective capacity’ for the array. Also, an array composed entirely of solid-state storage can reduce the amount of DRAM required for each application server, bringing down total cost in that way as well.


Another legacy concern expressed by IT managers and storage buyers is reliability. The notion that flash storage is less reliable than spinning disk is still out there. Ironically, the opposite is probably more accurate. Reports from Google several years ago and more recently from cloud backup provider Backblaze show most of the disk drives used by these companies have failure rates between 4% and 8% annually and a lifespan of around 3 years.

Flash has no moving parts, the primary culprit in disk drive failures, and flash has built into it mechanisms to address data reliability at the cell level. Since the NAND flash substrate degrades in minute amounts with use, flash controllers have always had processes in them to maintain data integrity throughout the usable life for the product.

This controller technology has improved dramatically over the past few years to the point where formerly consumer-grade MLC flash is now becoming the standard in enterprise storage products. Most flash vendors now build their enterprise SSDs for 5 years of continuous use, which puts them ahead of the < 3 year lifespan mentioned above.

Flash used to just be a ‘fast LUN’ that provided an immediate performance boost for critical applications. Now flash is a fast, reliable, economical storage system that’s becoming more common in enterprise data centers.  For more information on all-flash arrays and how they’re making this transition, tune in to the Storage Swiss on-demand webinar “Overcoming the Roadblocks to the All-flash Data Center”.

All On-Demand viewers, by registering once, will be able to access Storage Switzerland’s extensive library of on-demand webinars, many with exclusive white papers, without having to re-register.

Watch On-Demand

Watch On-Demand

Nimbus Data is a client of Storage Switzerland

Eric is an Analyst with Storage Switzerland and has over 25 years experience in high-technology industries. He’s held technical, management and marketing positions in the computer storage, instrumentation, digital imaging and test equipment fields. He has spent the past 15 years in the data storage field, with storage hardware manufacturers and as a national storage integrator, designing and implementing open systems storage solutions for companies in the Western United States.  Eric earned degrees in electrical/computer engineering from the University of Colorado and marketing from California State University, Humboldt.  He and his wife live in Colorado and have twins in college.

Tagged with: , , , , , , , ,
Posted in Blog

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 21,785 other followers
Blog Stats
%d bloggers like this: