Enterprise SSD means “Strict Standards Demanded”

IDC just released a new whitepaper (sponsored by Seagate) called “Storage Device Reliability and Endurance” by John Rydning and Jeff Janukowicz.

The title does not do this paper justice, as its focus in almost entirely on enterprise SSD.  If you are a techie and would like to learn more about the inner workings of NAND technology, you should check it out.  The paper does a great job of providing the background for what the root of the story is… enterprise SSD is here, but until the industry as a whole sets standards for measuring and marketing reliability and endurance, we may be faced with a lot of unhappy customers.

To quote the paper’s authors, “as the installed base grows and the market matures, the reliability of SSDs will be exposed — for good or for bad. Hence, it is most beneficial for device and system OEMs (for the sake of end users) to define a common set of metrics that characterize solid state storage reliability consistently and appropriately and to establish these definitions sooner rather than later.”

This is without a doubt the case…remember before the times of enterprise SATA drives like Constellation ES? Enterprise OEMs’ customers data was exploding, as budgets were remaining flat or in some cases decreasing, so what did many OEMs start doing?  Putting desktop and notebook drives in servers. Of course, these classes of drives aren’t designed for the rigors of enterprise computing, so obviously, they failed and the OEMs had a lot of upset customers.  The paper even points to this example as a case for lessons learned when considering SSD for enterprise storage.

So, when it comes to the adoption and growth of enterprise SSD as a storage medium in datacenters across the globe… some strict standards are demanded.

Agree / Disagree?  Chime in.

Related Posts

The Enterprise Perspective from the Flash Memory Summit 2010
Seagate Pulsar: the first enterprise-ready SSD
A tech fable: Flash and the Three SSD Applications
How SSD will emerge from the Trough of Disillusionment


2010-08-31T11:27:04+00:00

About the Author:

4 Comments

  1. Teresa Worth September 1, 2010 at 8:30 am - Reply

    I agree 100% about the need for enterprise SSD endurance standards. But it’s more than just standards – SSD products must also be designed from the ground up to meet the rigorous demands of enterprise workloads. The standards help define workload characteristics and enable customers to measure enterprise SSD endurance consistently, compare results across vendor products, and make better informed storage decisions.

  2. Matt Watson September 1, 2010 at 4:59 pm - Reply

    Seems like cheap SSD drives with over 1 million hour endurance ratings are good enough at this point?

  3. Teresa Worth September 2, 2010 at 11:32 am - Reply

    Is a cheap SSD with 1M hours good enough for enterprise? Probably not.

    It is important to not confuse MTBF (Mean Time Between Failures) with endurance. Take this analogy: Tires typically do not fail due to how long you own them. They wear out based on the distance driven and the conditions under which they are driven. The same is true of SSDs and this endurance factor is not factored in published MTBF numbers. An inexpensive tire used for the family car may last 20,000 miles, but if that same tire was used in the INDY 500, it probably would not last more than a few laps. The same would be true of an inexpensive client SSD used in an enterprise application.

    Enterprise class and client class SSDs are designed for difference usage models. JEDEC Committee JC-64 defines and proposes standards for solid state flash endurance, quality, reliability, and durability methodologies and procedures (among other things – see http://www.jedec.org/committees/jc-64). JEDEC recommends that endurance ratings be based on a use scenario in which the SSD is:

    1. Actively used for a recommended period of time (data is continuously read from and written to the entire device) then

    2. Powered-down for a specified period of time during which the data must be retained.

    What type of workloads and testing methodologies were used by the “cheap SSD drive” to derive the 1M hour endurance claim? At what temperatures? What was the data retention rate for the device and over what time frame? How does this compare to the the workloads and usage model within your data center? How important is the data you want to store on the device? What happens if the device fails?

    Until SSD standards are adopted industry wide and vendor claims independently verified, it is very difficult to accurately judge or compare vendor claims on product endurance and reliability. It is better to be cautious than sorry when it comes to protecting your enterprise data.

  4. […] Enterprise SSD means “Strict Standards Demanded” Dare I say there’s too much data What’s a FIPS anyway? Cliff Notes: Seagate whitepaper on cloud computing Mommy, where do hard drives come from? Intel Atom processor’s impact on storage Momentus FDE Drives Achieve FIPS 140-2 Level 2 Certification Music’s new metric: bytes NSA conference attendees want the whole SED package […]

Leave A Comment