*For primary storage that serves workloads appropriate for all-flash, assuming you care about primary data storage systems at all.
In other words, we’re almost done talking about flash. Wait, what? When something becomes ubiquitous, it becomes uninteresting to talk about. Over the past couple of years, there was tension between “all-flash” and “not all-flash” arrays. Customers have said, “I can only buy an ‘all-flash’ array because everything else is ‘obsolete’.” Storage systems were purchased almost solely because of the media they contain. When every array becomes “all flash”, however, the conversation can expand beyond media. Simply being “all flash” isn’t a compelling argument anymore.
Before we shift the conversation away from flash-centric topics like V-NAND, TLC, floating gate transistors, internal charge pumps, and hot electron injection, however, let’s understand how important flash has been to storage and IT infrastructure. (Besides – hot electron injection? Every conversation should start with hot electron injection. I feel like The Rock may use hot electron injections…)
First, we’ll talk about how flash disrupts storage media but not storage systems. Second, we’ll cover the evolutionary impact of flash on storage systems. Finally, we’ll talk about how flash can be a catalyst for the disruption of IT infrastructure. (And just for fun – hot electron injection!)
Storage Media – The Disruption
Flash is a disruptive technology… to disk drives. With space efficiency techniques, flash is already more cost effective than disk for primary workloads. While deduplication and compression also work on disk, the latency and performance impact can be significant. Flash’s dramatically better random-access performance makes it a better fit for combining space efficiency and primary workloads. As SSD capacity rapidly expands, the gap is widening quickly. Flash is already a better media technology for traditional primary workloads, and the advantage is only expanding.
This shift is reminiscent of what high-capacity disks have done to tape over the last decade. With space efficiency techniques, disk became more cost effective than tape for traditional backup. The space efficiency techniques were possible because of disk’s dramatically better random-access performance vs. tape. As HDD capacity rapidly expanded, the gap widened quickly. Disk became a better media technology for traditional backup, and the obvious advantages are revealed in the revenue numbers.
Over time, media will shift again. Flash will displace disk as protection storage (especially as we move to copy data management). New memory technologies will displace flash for primary I/O performance. Each time, we’ll marvel at the performance, capacity, and cost shifts. Each time, the hardware market will shift. Software, in response, will shift with it.
For now, though, flash displacing disk drives is enough to deal with.
Storage Systems – The Non-Disruption
Flash is not disrupting the primary storage array market. Disruption happens when there are new markets with new value networks. Customers change technology, people, and processes. They solve existing problems in different ways, with completely different approaches and different people. Existing vendors find themselves blindsided by the disruption. They don’t invest in the disruptive technology, miss the market, and find themselves out of business.
There is no sign of a flash-induced storage industry disruption. All-flash arrays are not changing customer behavior. Customers buy and deploy all-flash storage arrays for the same reasons and workflows that they previously bought disk and hybrid arrays. They’re bought and run by the same IT groups. The broad trends demonstrate that the leading vendors remain strong. EMC has, by a wide margin, invested the most in flash storage (in excess of $3Billion). EMC, the overall storage leader, is the All-Flash Array leader. The vendors who led in disk and hybrid storage continue to invest and address their customer needs in the All-Flash space, as well.
This lack of secondary industry disruption (beyond the media itself) is not unique to the flash conversion. Following the “disk backup” analogy, backup appliances delivered huge evolutionary value, but didn’t revolutionize the backup market. The leading software products in 2001 – Veritas NetBackup, EMC NetWorker, and IBM TSM – are the major products today. Like the storage array vendors, they each invested significant effort to better leverage disk as target storage. Disk has replaced tape as the backup media, but customers continue to buy the same backup software and run it in the same way. The tape market was disrupted, not the backup market.
Thus, when vendors say – “Media X changes EVERYTHING”, they may be exaggerating. Sometimes “Media X” changes … just the media.
Flash is disrupting the storage media market, which has led to a significant evolutionary investment and innovation in the storage system market. Like disk backup, the disruption doesn’t go beyond the media, but the media change is driving considerable investment in innovation.
Flash is enabling storage system companies and customers to evolve and solve problems we’ve stuggled with for years. In the next post, we’ll talk about how flash has changed how we design storage arrays, build business models, and enable customers to do more with their storage.
Welcome to the Year of All Flash. It should enable us to talk practically about how important flash is to our industry, without needing to resort to hyperbole. (And we get to talk about hot electron injection!)
Stephen Manley @makitadremel