Mark Adams

All-Flash Data Centers, Part 2

Blog Post created by Mark Adams on Jul 24, 2016

The Register broke some very interesting news late last week by citing three independent sources that EMC will stop investing in further development of their XtremIO product line and put it in “maintenance mode”. EMC has denied this report but it does call into question how well today’s all-flash arrays will be able to transition to the coming all-flash datacenter which I covered in my last blog.  SSDs will soon be available in much larger capacities than can be found today and, according to the Register’s sources, the architecture of XtremIO isn’t capable of addressing the increased metadata sizes for managing the larger data de-duplication and compression tables that come with higher capacity SSDs. 

 

AFDC.png


About 70% of the all-flash arrays deployed by enterprise customers last year had less than 50TB of raw flash capacity.  That’s just a small fraction of the total storage requirements that enterprises install in a year and indicates much more flash capacity is going into tried and true enterprise systems.  These are what some of our competitors derisively call “legacy systems”.  But newer all-flash storage architectures haven’t proven that they can scale to meet enterprise capacities nor provide the advanced data services that enterprise customers require. 

 

Transitioning to an all-flash data center requires a lot more than what’s being offered in today’s all-flash arrays.  At HDS we believe we make the best flash solutions with Hitachi Accelerated Flash for our Virtual Storage Platform G and F series models.  Our custom built Flash Modules are superior to SSDs as they off-load a number of flash related functions like wear leveling, block/page mapping, write buffering, data compression and more.  We move the running of these flash functions from the storage controllers down to the quad-core processors that are on every Flash Module in a VSP G or F series system.  This is far more efficient and alleviates IO bottlenecks that will occur with other all-flash array designs when those systems are placed under loads. 

 

But our superior flash hardware isn’t the end of our value proposition. It’s the advanced data services in our software that also sets Hitachi apart.   Examples of this include Global Active Device for active-active clusters ensures continuous operations even in the event of a system failure or site outage. Hitachi Automation Director automates common and repeatable management tasks. Hitachi Infrastructure Analytics Advisor monitors application performance against SLAs and identifies resource constraints across your entire infrastructure.  And we back our flash solutions up with our data reduction guarantee and the industry’s only data availability guarantee.

 

COOP.png

 

Our customers trust Hitachi to help them in their Digital Transformation to an all-flash data center so they can better deliver IT-as-a-service.  In our last fiscal year (April 2015 – March 2016), we had 56 customers deploy at least a half-petabyte of our flash.  One customer, COOP, is  a Swiss retailer that is running their business on VSP G1000’s with our Flash Modules.  COOP reports that they are seeing consistently fast response times (<1 millisec) and producing much better analytics about their business operations. As a result, they’ve seen a significant reduction in their inventory costs through more efficient stocking of their stores.  Best of all, they were able to migrate all of their mission critical applications to their new flash VSP G1000’s during business hours and without any downtime at all. How many of the new AFA architectures out there are capable of that?

Outcomes