kentoh - Fotolia
Tracking storage technologies is a lot like riding a roller coaster -- some techs are so cool it looks as if the sky's the limit for their application potential, but almost as soon as you get all jazzed up, you find yourself on a white-knuckle plunge back to reality as mundane details start to mount up. Suddenly, tedious gotchas such as cost, compatibility, durability and integration make the technology look more like a science project than the next big thing poised to revolutionize storage shops.
Compounding the frustration of tech watchers is the traditionally slow pace of storage system evolution. In most shops, storage is a big, well-entrenched and foundational cornerstone in which even small changes can become monumental undertakings. In IT, technology inflection points tend to happen when engineering, science and R&D efforts coalesce into something that's practical and often with game-changing potential. But with poky storage just lumbering along, those changes are often so slow to unfold that it's hard to pinpoint whatever milestones they may be marking.
In this decade, arguably the biggest thing to hit storage has been the emergence of flash, which has bucked the usual creaky progress of storage techs by getting on the radar of just about every storage manager in record time. And now I can see several trends or indicators that suggest solid-state has arrived at one of those archetypal inflection points.
There are three discernible changes that support my current thesis, involving where we're now focusing our attention on solid-state technologies, deployments and implementations.
Focus shifts from chips to systems
Not very long ago, much of the solid-state discussion was about chip-level or related technologies. Early flash shoppers had to choose between single-level cell (SLC) and multi-level cell (MLC)-based devices, weighing speed and durability against price and capacity. Vendors sought to differentiate their products with specially designed ASICs and firmware that helped to lengthen the useful lives of solid-state products.
Though relatively recent, those conversations seem a little rigid today, like something out of the distant past. That's because flash vendors have made such tremendous strides in chipset and supporting technologies. The SLC vs. MLC debate has subsided and vendors routinely entice buyers with long-term warranties on their flash storage. Storage buyers today are shifting their attention to the system level -- how solid-state technology is integrated and used, and discussions about ROI and TCO factors that include such traditional core considerations as data center footprint and power consumption.
More flash alternatives, closing the gap with RAM
Another inflection point indicator is the sheer diversity and range of solid-state options. Flash storage initially found its way into enterprises by mimicking the familiar form factor of SAS or SATA spinning disk devices. Fusion-io pioneered PCI Express (PCIe)-based solid-state storage, which has proved popular for pumping up the performance of a single physical server. That success fueled a new flash debate over disk form factor vs. PCIe bus implementations and shared vs. dedicated flash resources.
Those points of contention have mostly faded away as the number of possible solid-state storage deployment alternatives has grown -- SAS/SATA DAS, PCIe, NVM Express, hybrid arrays, all-flash arrays, dual inline memory module (DIMM)-based persistent storage and now non-volatile DIMM (NVDIMM). The software side of solid-state solutions has been equally active, with advanced automated tiering, caching, dedupe/compression and flash pooling apps adding considerable value to still relatively pricey hardware. And among the more exciting recent developments are techs like the aforementioned NVDIMM that bring persistent storage closer to memory, thus serving as a preview of the inevitable blurring of the line between memory and storage.
It's all about apps
Perhaps the most telling development that implies a turning point for solid-state storage is how the spotlight has gradually swung from solid-state's hardware and software factors to its practical applications. Vendors and users alike are more far more likely to approach a solid-state purchase today with specific application performance goals in mind. Sure, that was essentially true when flash first hit the market, but now the performance considerations are more sophisticated, reflecting the diverse alternatives available today. In addition, our ongoing research finds that more than one-third of companies have implemented some amount of flash somewhere in their storage environments. Initial forays were typically modest, but have served as proof-of-concept deployments that have paved the way for the wider use of solid-state storage. Just a couple of years ago, flash storage was typically acquired to address specific performance issues for one or two mission-critical applications; today, users are looking at more general-purpose purchases where the benefits of flash can be spread to more applications.
And the next inflection point ...
Enterprise-ready solid-state has come a long way in a short time, but most industry watchers are still anticipating the new inflection points the technology will encounter. Maybe those watershed moments will come when new techniques like 3D architectures push the cost of flash below that of hard disk drives. Or maybe some of the new solid-state techs like magnetoresistive random access memory (MRAM) or Memristor that are beginning to emerge in real products will overtake NAND flash and usher in a new stage of solid-state evolution. I don't know which will come first, but I'm betting that both will occur -- and sooner than we think.
About the author:
Rich Castagna is TechTarget's VP of Editorial/Storage Media Group.