 What's the history of flash? Well, the early days of solid state as an enterprise storage device really started in the 1980s with, believe it or not, mainframes. Certain vendors such as STC, National Advanced Systems, and EMC competed with IBM Direct Access Storage Devices by packaging RAM, RAM memory actually into large storage devices. The advantage of these systems is they were much faster than spinning disk because they were, of course, RAM based, but the major disadvantages were they were volatile, i.e., if power got cut, data was lost, and they were really, really expensive. They had to be backed up with batteries and never really went mainstream. As such, they were niche oriented and confined to fringe applications where IO speed was really critical. It wasn't until the mid 2000s when Apple replaced the iPods spinning disk drive with an all-flash music player. Remember that? This created the volume necessary to jump start. What we look at at Wikibon is the modern flash era. The big difference is that flash, unlike earlier solid state disk drives, which were based on RAM, is a persistent media, meaning like magnetic disk, it retains data without a power source. Apple's commitment to large scale production drove prices down and fueled adoption in consumer products. Now, in 2008, EMC and Estec partnered to deliver what Wikibon called at the time a haymaker to the enterprise storage market by packaging enterprise SSDs into large scale storage arrays. This kickstarted the modern era of enterprise flash storage and many dozens of companies began securing VC funding to attack this new market. Over the next four years, we saw innovations in flash, ranging from PCIe connected flash devices to flash as an array cache to all flash arrays to hybrid flash and spinning disk systems and eventually flash that extended main memory as a directly addressable resource. Now in 2012, we're seeing the extension of the storage hierarchy where flash is being managed from the server all the way through the spinning disk array. Now, the key innovation here is software that intelligently places data at the most appropriate location in this new emerging storage hierarchy. What's the vision of where flash is headed and how is it changing enterprise storage? At Wikibon, we see flash becoming the predominant medium for IO intensive applications. Because it's more expensive than SATA disk, flash will initially remain focused only on data with high IO activity. However, because data continues to grow at 50% per annum or more in spinning disk performance is mechanically constrained and as a result, not improving, the effective cost per IO of flash is becoming much more attractive than that of spinning disk. What this means is that while historically, cost per gigabyte was the primary metric used by practitioners to evaluate business cases for storage, increasingly cost per IO is the more meaningful comparative statistic. Specifically, the flash cost per IO is declining dramatically while the cost per IO of spinning disk is essentially static. Spinning disk will remain the cheapest medium for long-term storage. This means the role of spinning disk over time will become increasingly focused on storing data with less frequency of access, for example, archiving, while the majority of active data will eventually land in the flash hierarchy. Now, there are two important points to remember here that are relevant. One, this transition is going to take some time, perhaps as much as five years. And two, software to optimally place data in the hierarchy will become more and more important. Now, as well, mechanisms to provide data consistency, cash coherency and data sharing will over time become crucial. Now, the final point is flash eventually is going to enable a whole new breed of applications that were once too expensive to justify based on the IO economics of spinning disk. For example, the ability to collect operational transactions, metadata, social media data, and multiple other sources of external data simultaneously in concert with operational databases will enable a whole new breed of analytics that allow companies to target buyers in near real time, just as buying decisions are being made. What does all this mean to customers? The key issue for most practitioners that we talk to in the Wikibon community is how can I take advantage of the trends in flash without ripping and replacing my existing infrastructure? There are three factors that customers should consider when planning for this future. Number one, remember over time, spinning disk will no longer be the most cost effective medium for active data that will migrate to flash. Number two, you will not be able to manually ensure the right data is on the right device. The industry must automate this data placement. Number three, start looking for business opportunities where applications can exploit the declining cost of IO and begin to monetize those applications in ways that were never possible with spinning disk.