Is the storage array on the endangered species list?
A high-stakes game is playing out today as Amazon, Google, and Microsoft compete for leadership in cloud services markets that are projected to total in the hundreds of billions of dollars by 2020. In the last quarter alone, they spent a combined $9B to build out data centers to support the exploding cloud market (WSJ, 4/7/17).
There is little question whether this triumvirate will be successful in their cloud efforts — they already are. All three have leadership momentum, technical competencies to drive efficiency, massive scale, no incumbent baggage, and strong balance sheets.
In comparison, today’s storage system companies are seriously disadvantaged in this market and I recently learned why. Citi analysts, including Walter Pritchard, released a report earlier this month entitled “Quantifying the Cloud” in which they pointed to two major economic drivers for massive change in IT: deflation and demand elasticity.
The Citi team of analysts evaluated 8 categories of IT spend (Labor, Communication Services, IT Services, Software, Servers, Storage, Networking, and UC Systems) for % cloud exposure, cloud deflation level, and demand elasticity. Storage ranked as the most negatively impacted in each category!
Let’s look at each of these:
- % Cloud Exposed — Citi estimates 100% of storage is exposed to cloud-driven deflation. Enterprises can choose to go to an offsite cloud or deploy Software Defined Storage (SDS) on premises to achieve economics approaching those of public cloud.
- Cloud Deflation Levels — Citi estimates cloud will result in price reductions of 65-95% for storage caused by use of open SDS running on commodity DAS or white boxes, and achieving higher data density vs. on-premises appliances. And, cloud vendors are subsidizing storage costs with more differentiated offerings.
- Price Elasticity of Demand — Citi calculated elasticity (defined as % annual change in units divided by % change in unit price) for storage over the last 5-year period and saw .78x Price Elasticity of Demand, meaning unit volumes decreased even with price reductions.
See the Citi graphic below comparing the elasticity of demand for storage to other product categories:
In short, Citi found storage hardware to be the most negatively impacted IT area by cloud growth. While IaaS, PaaS, and SaaS will grow at a CAGR of 17-21% through 2025, storage revenue will decline by -4% CAGR. Data is exploding, but incumbent storage hardware vendors will not participate in the market. Instead they will be left to fight to hold share in a very price-sensitive and declining market.
Storage isn’t going away, but it is going to the cloud, and on-premises arrays are a declining market segment facing severe downward pricing pressure (see above). Citi predicts the on-premise market will decline from $27B in 2015 to $19B in 2025 — a 30% decline — and it is clear that margins will also erode during this period. Products and business models must evolve rapidly to survive in a smaller, lower-margin market. Both on-premise and cloud vendors are now racing to deploy data reduction technologies to achieve “lowest possible cost” products by leveraging SDS, data compression, deduplication, and white box hardware. Citi projects that the lower prices and market share shift will continue. We think this accelerates the path to highly efficient SDS incorporating data reduction on commodity-class hardware for all on-premises and cloud storage.
Within SDS, we are seeing large organizations demand data management functionality (traditionally the domain of enterprise storage systems) that operates at the OS level. Virtualization and containerization trends allow software running in the OS to be easily migrated from on-premises to the cloud. We address this opportunity with our Permabit VDO for Data Centers data reduction software. Data reduction was initially an exclusive feature of storage vendors, and then it moved to proprietary SDS stacks. Today, VDO data reduction software is readily available as an OS-level driver for Linux, where it can be universally consumed.
VDO inline deduplication and compression can be deployed with SDS and commodity hardware to deliver storage at a fraction of the cost of traditional enterprise offerings. Also, since VDO is deployed at the OS level, it can be readily leveraged on-premises or in the cloud. VDO pricing starts at $199 annual for support and can present up to 4 PB of logical storage!
What will data center storage look like in 2025? Citi says it will leverage lower-margin, lower-cost on-premises and cloud storage. We think that shift will cause critical functionality to move to the OS-level. Just as VDO provides compression and deduplication to the Linux OS today, look for additional data center storage features (traditionally available from storage vendors only) to migrate to the OS and interoperate in a hybrid cloud world.