Currently Being Moderated

1 book reader.JPGAs the picture to the left illustrates, you’re never too young to learn about data storage.  In my book Evolution of the Storage Brain, I boldly made 37 predictions about the future of data storage. I looked at short term trends as well as trends all the way out to the year 2040.  When the book was published in early 2010, I promised myself that I would return to my predictions after a few years and see how they were standing up.  Well here we are, 3 years later, and I find that 10 of my predictions had to do with Flash and other solid state storage techniques.  Can’t think of a better time to review, so here goes…

 

2010: The solid-state folks will keep producing higher densities with lower costs, largely driven by consumer product requirements.

2013 Update: eMLC and TLC Flash formats have emerged, with many other upstart memory technologies looming on the horizon.  More about these new technologies below.

 

2010: Postage stamp-sized memory modules will begin to demonstrate ridiculously high storage capacities.

2013 Update: I am not sure if postage stamps will exist in the future, but it seems there will always be postage-stamp sized memory modules.  In 2010, CompactFlash SDHC cards were 32GB, whereas today’s SDXC cards are 128GB.  Not yet ridiculously high capacities, but getting there.

 

2010: From 2020 through 2030, a period of hybrid design will ensue as a bridge between the magnetic and solid-state eras.

2013 Update: Hybrid designs are emerging sooner than expected, using a combination of PCI Flash/SSD and HDD.  In 2013 it appears that nearly every enterprise storage system sold will have some combination of Flash and HDD.

 

2010: During this period, look for more attention to be paid to hybrid disk drives with large solid-state buffers.

2013 Update: Despite Seagate’s fervent promotion of the Momentus XT hybrid drive, there aren’t any takers in the enterprise space.  At this point, if hybrid HDDs haven’t caught on in the data center, they probably never will.

 

2010: Cached data will make the prospect of drive “spin-down” (and greatly reduced electric bills) a practical reality within 10 years.

2013 Update: Spin down is still a viable way to reduce energy costs, but IT is facing too many other issues to make this a priority.  Give this a few more years…

 

2010: Using vast quantities of read and write cache, storage controllers will, in fact, move data at the speed of light.

2013 Update: Vast quantities of cache have not arrived yet.  Modern storage systems typically consist of 5%-10% Flash capacity.

 

2010: Memory (read cache, write cache, hybrid disk drives and solid state disk) will grow faster than any other storage component in the immediate future. This will be one of the more interesting aspects of storage evolution for the next five years.

2013 Update: Well Flash certainly has become interesting, hasn't it? It is also attracting the interest of venture capitalists.  In a blog written earlier this year, I identified 6 startup companies offering nothing but all-Flash arrays.  It will be interesting to watch Flash technologies continue to evolve over the next couple years.

 

2010: Within 20 years, economies of scale will have driven memory prices down to levels comparable to disk.

2013 Update: In 2010, SSD capacity was 10 times more expensive than HDD.  As we enter 2013, SSD capacity is only 6.5 times more expensive than HDD...a trend surely to continue?

 

2010: Technologists will also have solved the persistence issue of flash memory.

2013 Update:  It looks like the persistence issue of Flash will only be solved by entirely replacing it with another technology.  Phase Change Memory (Micron), Spin Torque Transfer (IBM), and Memristors (HP) are the leading contenders to replace Flash.

 

2010: More dense memory technologies will continue to appear.  By 2040, all magnetic storage will have been replaced by solid state memory devices.

2013 Update: In addition to the 3 emerging technologies mentioned above, the idea of carbon nanotube-based memory modules has also surfaced, claiming to have “expected-unlimited” endurance.  The clock is ticking on both Flash and HDDs, but will keep ticking for many years to come.

 

So all in all, Flash seems to be moving forward pretty much as predicted 3 years ago.  Oh, one final prediction: by 2015 the clear successor to Flash will have emerged, to the dismay of users and the delight of storage vendors.  After a few more years of transition, Flash will join the ranks of floppy drives as a gone-but-not-forgotten chapter in our industry evolution.

 

Data Storage Matters,

 

Larry

Comments