Bob Madaio

Flash Trees, Forests and Shipments

Blog Post created by Bob Madaio on Nov 5, 2014

It’s that time again. Time to talk flash petabytes.

 

And maybe for the last time, at least as with this level of frequency and in this manner. If you are new to the HDS Community, you've missed that over the last few quarters, I've been blogging about specific HDS petabyte shipment capacities on my blog.

 

Why?

 

Among other reasons, there was a lot of “noise” in the industry from start-ups and market(ing) leaders on their respective progress with flash storage.

 

There were grandiose claims of leadership. There were very public, and very meaningless guaranties.

 

Hitachi, despite goofy soccer videos on Flash storage, tends to be a bit quieter, but we felt our flash progress was worthy of discussion.  So we/I did discuss it… 19.5 PB shipped a couple quarters ago (calendar 4th quarter of 2013), then ~18 PB (calendar 1st quarter 2014), which brings us to now.

 

In the 2nd calendar quarter of 2014, Hitachi Data Systems shipped just over 26.5 PB of flash capacity.

mic.jpg

(Well that was fun, but I'll pick the mic back up...)

 

Of note, over 24 PB of this was our unique Hitachi Accelerated Flash storage devices.

 

Most of this was within tiered systems, though this quarter saw some large all flash configurations of our massively-performing Virtual Storage Platform (VSP) G1000.  With performance easily equivalent to multiple individual systems from the start up “leaders,” such a system brings a whole new level of consolidation and data center acceleration.

 

So why might I move away from this quarterly trend... I mean, it's been all good news, right?

 

Well, for one, I think the point is made.  (Sorry about the mic-dropping.)  We are a significant player in this space.


And whether we are the biggest or not, the real point is that Hitachi has figured out a unique solutions-led approach to flash that customers have responded to. Shipment capacity counting just happened to be one simple measure of that customer and market response.

 

Another reason to move away from this quarterly fixation on capacity counts, however, is that the market is maturing enough to look at flash more holistically. To continue to call it out as a standalone discussion seems misleading.


To leverage an old but familiar refrain, customers have started the move from the proverbial “tree” discussion, to (re)focus more on the “forest.”

forest.jpg

 

I’ve talked about this before (like in this HDS Webech that I did with Evaluator Group’s Randy Kerns), but flash is not a solution.  Flash is a component to system, which itself is only component to a full solution.

 

Having just finished up a week of customer meetings in New York City and being in the middle of a visit in the Nordics with our customers and our local teams, I've seen a change in the conversations we are having.

 

And yes, there are still some customers looking at “point” technology questions around flash.  What’s the BTU?  How many KVA? What’s the MTBF? Etc.

 

But the majority are seeing the bigger picture. They have realized that focusing on a single flash system's specifics is less important than focusing on the overall solution and seamless acceleration of key business applications.

 

One customer installed our Flash in a tiered system without having told the database team what they were doing. The database team called "immediately."  Why? They wanted to know what happened, and to proactively thank the infrastructure team, claiming it "was the best thing the IT team had done for them" in recent memory.  What mattered was the database performance - not the flash installation.

 

Another customer said they don't understand the value of single-purpose flash products for customers that have a solid infrastructure management plan in place.  Instead, they wanted to focus on having the ability to add workloads into flash tiers or pooled hybrid tiers, keeping management simple. They understood that certain highly "dedupable" workloads on such systems could provide a tactical benefit, but that it hardly seemed worth the disruption to their environment given the overall plan they for managing tiers, data protection and high availability.

 

What I took away from these (and other) conversations, was this: whether our terminology of Continuous Cloud Infrastructure has fully resonated in the our customer base or not is debatable, but the concept is quite real. Customers want a solid infrastructure, unified management and a way to bring new applications into the mix quickly and flexibly. And yes, they also want the performance benefits of flash, and they expect it to fit within their established plans and to leverage the flash performance benefit and investment across applications.

 

So yes, we are shipping a quite a lot of flash storage. Maybe we are seeing such success simply because our customers think we build the most reliable, highest performing and completely trusted storage they can get their hands on.

G1000.jpg

 

Maybe. But, I don’t think that’s it. Not alone, anyway.

 

Nor is it that we write the best and most blogs. Or, I'm afraid, simply because we are the best people to work with. Though, we try to be.

 

I think customers with an enterprise mindset are maturing. And when it comes to flash, we are seeing less tree-huggers and more forest cultivators who are more interested in the big picture. And that is exactly the mindset that our incredibly powerful, but surprisingly simple, flash strategy aligns with quite nicely.

Outcomes