The headline of the latest IDC storage report says a lot:
Weak High-End Demand Results in Worldwide External Disk Storage Systems Revenue Falling at Rates Not Seen Since 2009, According to IDC
A tweet from Dave Reinsel, a Group VP at tech analyst firm IDC was even more evocative:
Three years pass and twice as much capacity is shipped at the same revenue. This is not just Kryder’s Law of disk density at work, this is an inflection point in the type of storage being purchased.
The key phrase is “weak high-end demand.” High-end storage has been dominated by three players for the last decade: EMC (Symmetrix), HDS (USP, VSP), and IBM (DS8000). Of the three, only EMC publicly breaks out financial information about the high-end. It immediately made me look back at some of the slides I had seen in some of the latest EMC earnings presentation, and wow:
A look back over a couple of years show flattening growth and then a gradual decline.
This is a big deal in the storage world, and reflective of a broader shift in data center design beyond EMC’s results and weakness in the High-end. The environment has changed dramatically:
1) 80% of applications are virtualized, which has workload implications, and an increasing amount of apps, particularly Cloud apps, leverage non-persistent and parallel interface patterns
2) The amount of data has grown within and across data centers, with more types of data and ways to mine it
3) x86 hardware has matured to the point where software can effectively subsume all relevant storage capabilities
4) Flash memory introduced exponential improvements into the data flow, requiring new designs (still in flux as nicely captured by Chris Weis)
These factors have introduced competing paradigms and products to the traditional High-end, and traditional SAN and NAS architectures overall. Theories like the two-tier storage architecture with low-latency server-based storage and unified object-based storage working together (first expounded upon by Richard McDougall of VMware in 2010), and “server-based SAN,” are no longer theory, but in production in large environments. Silos of traditional High-end, transaction-optimized storage, are increasingly being replaced or sidelined by converged systems, hyperconverged systems, and flash arrays, driven by new storage software stacks.
The incumbents are not standing still. They are actively attempting to evolve or disrupt themselves. I predict a significant shake out in the next 9-12 months.