What Really Drives Cost Efficiency in Storage?

As I write this blog, I am reminded of this quote from a McKinsey report on the semi conductor market, “if the automotive industry had achieved similar improvements in performance in the last 30 years, a Rolls Royce would cost only $40 and could circle the globe eight times on one gallon of gas – with a top speed of 2.4 million miles per hour.” Sweet.

When IT folks look at cost efficiency they need start no further than their own racks for the best deal of all time. In the last 30 years Moore’s Law has delivered a 32,768x factor of improvement. Refresher – technically Moore’s Law says that the number of transistors in a chip doubles every two years, but it’s generally accepted that most figures of merit – performance, capacity, cost, etc. – also improve by a factor of two every two years. 32, 768 is just 215 – Moore’s Law improvements for 30 years.

If improvements in semiconductors were the only effect at play in the storage industry, the number of storage system vendors would be about the same as the number of semiconductor vendors because semiconductors and storage would compete on the same characteristics. But there are 71 all-flash array vendors today. Why? As Marc Andressen said, software is eating the storage world.

Storage has been a software problem, since the beginning of the modern era in the 1980’s. After all, what is RAID (Redundant Arrays of Independent Disks) other than a mathematical algorithm implemented in software running on a CPU? And even back in the day, storage engineers filed patents to add new capabilities to the basic data protection provided by RAID, in order to differentiate their products from one another.

So here we are, almost one-fifth of the way into the 21st century and the story continues — innovate, deliver and repeat — as semiconductor economics enable a new generation of software inventions.

Let’s look at Kaminario’s software architecture and how it delivers outstanding cost efficiency:

  • Inline data reduction. Puts flash storage in the cost ballpark of high-performance disks.
  • Scale up and out. Optimize spend for performance and capacity.
  • Adaptive block size. Consolidates multiple workloads to buy only one system.
  • Flash friendliness. Minimizing write wear on the SSDs enables less expensive SSDs and makes them last longer, for lower cost.
  • Perpetual array. Investment protection at its finest. Don’t throw it out. Just add to it.

The bottomline for cost efficiency: software techniques make the effective cost per capacity of an all-flash array the same or less than conventional, with up to a 100x performance improvement for the same price. Even though solid state drive (SSD) media is more expensive than high-performance hard drive media. Better cost efficiency no matter how you measure it.

Back to our Rolls – if we account for just additional Moore’s Law effects in the almost six years since the McKinsey quote, it would travel to the moon three times on a tank of gas at three percent of the speed of light.

What do you think?

New Call-to-action