FLASH! Will overtake Disk in 2016!

Well, almost. 

I think we all agree that Flash storage will become the standard in the future, even over slow disk.  The question is how soon?  

There is a well written research article in Wikibon that talks about architecture, pricing and planning for Flash.  Plenty of other sources as well that all point the same way. The chart above comes from that Wikibon article.  http://wikibon.org/wiki/v/Evolution_of_All-Flash_Array_Architectures

If you look at the chart, the projected 4 year $/TB (TCA) for Flash falls below the same $/TB (TCA) for Disk around mid-2016.  That trend not only continues but within 4 years is about 8x lower than Disk.  When Flash becomes less costly than Disk, the choice for storage technology is $imple. 

Will it happen quickly or slowly?  That's a good question and maybe this old disk sales chart will help answer that.  Going back many years, we can see that once the cost of Disk reached a certain threshold, the barrier to entry was knocked down and Disk sales exploded very quickly - much faster than the drop in $/MB of Disk. It simply became affordable for most companies.  I believe we will see the same explosion in Flash sales shortly after that cost drops below Disk. 

Flash is quickly becoming affordable.  As it does and storage teams begin consuming it, what do need to think about in your IT organization?

How does your storage organization change to support it? Single Tier or multiple Tiers?  Which Flash Architecture do you pick? Do you only pick one? What about backup now that you can actually use snapshot technology because reads aren't contending for disk arms? Does your storage monitoring software work with Flash and at a granular enough level to be useful?  Do you even need it?  Can I build more redundancy into my storage design as costs plummet?  Do I need the same number of staff and staff skill? 

What about your server teams?  How does their server capacity profile change when their utilization is suddenly driven up because systems aren't waiting (as much) on I/O?  Are they panicking or embracing this? As Flash becomes dominant, does the OS change to support a new I/O stack that can treat Flash as Flash and not as Disk?  Do my file system layouts change? Does my HA design change? 

What about the data center?  Does your Facility team understand you won't be needing all that floor space and power with Flash?  Sure, cooling will be less as well but the demand for it will be in a much smaller footprint and the data center may not be designed for that kind of cooling density.  What about that Generator and UPS upgrade?

How about Operations?  Onlines and batch will run better, sometimes much better. After the switch to Flash, what will the new signs of a problem be? Does application monitoring software handle Flash usefully? How about those hard to get scheduled downtime windows? Will Flash enable more of them or maybe new storage designs can eliminate them? 

How about your D/R provider?  How quickly are they getting on board with this Flash changeover?  Once you go Flash, you won't go back to Disk.  Can replicate data now or maybe ship Flash drives for recovery? 

Finance may have a say in this too.  Does depreciation change?  Are you refreshing more often or less often than today?  Maintenance cost planning will change for everything.  

Plenty to think about isn't there?  The technology is the easy part.  Make sure everyone else is on board.

You don't want to Flash anyone.