Are enterprise flash advantages sustainable?

Mon. Jun. 30, 2008

There's been a lot written about flash SSDs lately, starting with EMC following their announcement of the technology, and followed by other companies wanting a market position in relationship to flash.  From a strictly marketing perspective, EMC pulled off an excellent move, forcing other companies in the industry to respond.  That's what industry leadership looks like. 

Of course, SSDs have been around for a long time, but the big deal this time is flash memory - as opposed to dynamic RAM.  There have been attempts and thoughts about flash for enterprise storage for many years. I was at Convergenet when they attempted to make it work a decade ago.  They couldn't.  I recall speaking with Joan Wrabetz (Ex CEO of Tricord) several years ago and she was telling me about how the development of flash memory for consumer goods was going to have an inverse "trickle up"  impact in enterprise storage.  That's not to say that iPod flash is the same as enterprise flash, but it does help to have money flowing into an industry to stimulate technology advancements.

Barry Burke, the Storage Anarchist blogger from EMC has written a couple of good pieces recently.  But he wrote something today that caught my attention and made me pause and wonder: 

"And judging by the applications I've seen flash targeted for by customers, the early adopters are going to have a pretty big competitive advantage on their competitors."

Really?  Like all the technology advantages that Nicholas Carr was discussing a few years ago in his book Does IT Matter?  Even if EMC's customers can extend some advantages from EMC's flash products, how long will it take the rest of the industry to gain sufficient competency with flash technology and how much cheaper will it be for everybody to implement afterwards? I'm sure EMC has a road map, but road maps don't necessarily predict where the roads get bulldozed. Believe me, I'm not hinting at anything here, but I do think EMC will have its work cut out making its first mover advantage stand up over time and it could turn out to be one of those deals where being a first mover is actually a disadvantage because the technology is already so widely available.  It's a risky business for EMC - leading the way with a technology that is already sliding towards the great plateau of commoditization.

del.icio.us email this digg digg
Only published comments... Monday, June 30, 2008 04:57 PM
by Marc Farley, Inside IT Lead Blogger
Filed under:

Comments  Comment RSS Feed

Nik Simpson said:

I agree, this first generation of flash-based SSD is pretty simple for anybody to integrate, and I doubt EMC's first mover advantage will last 6 months. My bet is that most storage vendors will have introduced SSD by this time next year and that customers will have relatively widespread deployment for high performance apps by sometime in 2010.

The more interesting questions are:

  1. Impact on the high-performance disk market, I can't see much long term future for 15K SAS/FC drives, you only buy them for IOPs and flash-based SSD will kick their spinning behinds ;-)
  2. Scope for innovation in how we use flash. Today, the solutions are pretty unimaginative, just treat them as very fast disks. But long term you have to wonder if that's the best use. Why make them look disk when they are truly random access? If you designining a filesystem why bother trying to limit fragmentation? Why hang them off a bus designed for conventional disk using a protocol with similar limitations?

So lots of scope for interesting things to happen as flash-based SSD becomes mainstream.

June 30, 2008 9:33 PM
Marc Farley, Inside IT Lead Blogger said:

I think you pose the most interesting question Nik - what will the scope of flash applications be and how will expanding the scope change the communications methods used?  Flash might not be the most interesting technology to consider - there are likely going to be new memory technologies at some point, such as phase change memory (http://en.wikipedia.org/wiki/Phase-change_memory).  Will it still make sense to use conventional file systems with block storage abstractions?  If you assume random access memory is a storage device instead of spinning disks behind a constrained read-write channel, the entire universe could look quite different.

Of course the technology that has lagged the most over the last 25 years of technology development is filing (file systems mostly).  Its almost embarrassing that so little progress has been made compared to all other IT technologies. The reason why is that file systems have been the least common denominator, maintaining system and application compatibility.  When, if ever, will there be a chance for file systems to make a break for it?

July 1, 2008 7:23 AM

Leave a Comment

Compose
Preview
(required ) 
(required , not published) 
(optional )
(required ) 

Note: Conversation is encouraged and expected. However, moderation of comments is necessary to prevent spam, personal attacks, profanity, mentions of legal action or off-topic commentary. We will not publish comments that advertise third-party shopping sites or ones that violate our terms of service.

Comments related to specific product support or customer service issues will be addressed separately rather than posted here. Please use the links in Contact Us for product and customer service assistance.