Storagezilla

Lights out.

In a case of stating the obvious UCSD & Microsoft have decided that come 2024 Flash is a doomed storage technology. That won’t matter to too many of us as we’ll have started the move onto the next big thing by the end of the decade.

I’ve stuck my neck out on that as far as it’s willing to go without having to face the executioner but there’s more reasoning behind it than just the issues UCSD-Microsoft discuss in their paper. However, just how it is stated every year that Moore’s Law is three years away from hitting a wall the authors don’t take into account semi-conductor manufacturers coming up with smarter ways to work around Flash’s well known issues.

The bottom line is the higher the density of Flash the lower the performance and if it’s SLC or MLC will continue to matter now and well into the future.

As people are hoping to close the pricing gap between Flash and Disk Drives and putting aside the fact we’d need to spend about $100B building new Fabs just to meet that capacity requirement, the fact remains that with current technology the closer you get to a disk drive capacity price point the less of a random read benefit you’re getting from the Flash itself so you might as well just use a disk anyway.

Fast low capacity Flash devices will kill off fast low capacity disk devices. Expect the disk drive industry to turn towards dense and ultra dense technologies with improved reliability as they exit the high speed disk segment.

EMC ItemPoint

Item level recovery for MS Exchange from Snaps, Clones and Bookmarks.

Sam is…exuberant. I love the drop in at 7:05.

Demartek benchmarks VFCache

Dennis does a good job going through VFCache tech and shares some of his benchmark results.

Remember, unlike fictional products from lagging competitors it’s available from EMC Channel Partners & EMC Sales today.

Tearing down the wall of DAS

I just saw a rather ashen Dave Flynn field yet another question about FusionIO going on the auction block due to depressed margins.

The Street doesn’t sound convinced that FIO are going to continue as a stand alone company and they’re now positioning for a bidding war. (Has anyone seen Frank Quattrone today?)

Let me be completely clear, Dave doesn’t understand what depressed margins are yet.

The high roller days for FIO are over.

For the first time in its life it gets to say hello to competition and competition drives down prices. Not good if you’ve been booking losses in your recent earnings.

Now, FusionIO have some unusual things going on. When you consider that majority of their revenue comes from two customers gained at a time when there was no competition how can you be flat or down outside of those when the Flash market overall is expanding in orders of magnitude?

Look at their financials. They have not made real progress and aren’t growing at a consistent rate in what’s the largest technology expansion in storage business over the past five years.

Today they’re a DAS company. They have about 70% of the PCIe Flash DAS market. That’s their primary revenue stream and they’ve wedded themselves to the idea that low latency basic block storage is all you need. But we know DAS isn’t shareable. We know it isn’t scalable. We know it’s a management nightmare. We know data protection gets much more complex than if you were using shared storage.

But they’d like you to forget all that. The Cloud or the Supreme Being or Santa makes those issues irrelevant.

I see that 70% share as vulnerable and suspect it will dramatically erode now there is competition in their market.

To look at their software the Fusion play appears to be around saying your server CPUs can take the hit doing all the management and wear levelling of the Flash devices. They don’t implement standardised functions in the silicon like Micron do to offload those tasks from the server so the viability of a server consolidation play becomes questionable when you have their software burning through continuous cycles just trying to keep the storage functioning.

On the topic of software Fusion has two product offerings, one for physical environments directCache, the other, ioCache for virtual environments. It’s interesting to me that ioCache still only supports the ESX 4.x hypervisor and only for Windows 2K8 guests.

There’s obviously a re-write required to support vSphere 5 and who knows what is required if they want to munge directCache & ioCache together.

No matter how many people Dave has cold called since last week the fact remains that his part of the world changed today and the good times are over.

If you can’t make money when no one is competing against you life only gets harder when they are.

Flashpoint, with Pat Gelsinger.

I spoke with EMC President & COO Pat Gelsinger about the VFCache product launch announcement, which you can watch and ask questions about here, discussing the new products themselves, how EMC is doing in specific and it’s industry placement in general.

Take a look.

-----

VFCache. Cutting through the noise what’s the bottom line?

At its heart VFCache extends the value of the storage array directly into the server for ultimate application performance. It’s storage caching intelligence moved closer to the CPU.

For EMC arrays only?

While we’ll have deep integration in unique areas such as Fully Automated Storage Tiering (FAST) and systems management, VFCache read caching will be general purpose and work across multiple storage array providers in a Storage Area Network.

So customers could if they choose use VFCache in front of non-EMC storage? Interesting. To focus on ‘Split card’ functionality, where server DAS & SAN workloads can be accelerated at the same time, with larger capacity cards coming to market do we envision EMC Direct Attached Storage only deployments going forward?

Split card functionality is a great way for customers to use VFCache in both storage worlds, yes. Caching for extra performance with the protection offered by arrays for mission critical information and DAS for more ephemeral data sets such as temp files, swap and indices. We will provide larger capacity cards over time and the DAS use cases will naturally expand as a result.

What next for VFCache?

Even deeper FAST integration. We’ve done a lot of work in the field of I/O hints and how we can make the array and all the server caches themselves even more efficient. If you have VFCache in a server you could optimise array cache for those which do not and that has a knock on effect on things like promotion and demotion statistics in data placement.

All of that ties in with the next VFCache focus area, Distributed Cache Coherency. DCC maintains cache coherency between systems using VFCache allowing for the construction of Active/Active clusters or support scale out computing.

Those are two near term additions from a bunch we have coming next for VFCache.

Now Project Thunder..

A new product from EMC in a segment we’ve never been in. Server Networked Flash.

InfiniBand or Ethernet connectivity to Thunder Appliances where you scale up inside Thunder and scale out with additional Thunder Appliances. This is a server centric design providing Terabytes of Flash, tens of Gigabytes per second of bandwidth in support of millions IOPS with incredibly low latency at high loads. And there’s VFCache integration, we’re using the low latency RDMA data path assets developed for Project Lightning in Thunder.

You’re introducing Project Thunder at the VFCache launch will we see more at EMC World?

We’re considering that.

EMC innovation. Thunder on its way, VFCache today in 2012, VNXe & VMAXe in 2011, VMAX in 2009 and Atmos in 2008. What’s the innovation strategy going forward?

The strategy is an acceleration of the pace of organic innovation. We want to combine the best of breed inorganic innovation with best of breed organic innovation. The result will be unparalleled innovation, an acceleration of new product deliveries and an ever increasing gap between EMC and the rest of the industry in the areas of cloud, big data and security.

Is innovation something you can measure and set goals against?

There are some metrics I find helpful, percentage of R&D, number of patents and their impact and percentage of revenue from new products. However, while these are helpful, less tangible items like quality of talent, disruptive product development and university relationships are better leading indicators of a company’s innovation capability.

Between VMware, revenue leadership in Backup & Recovery from BRS and analytics from Greenplum is EMC still a storage company?

Going forward EMC wants to be the most disruptive data centre infrastructure company in our industry. While today storage is our centre and our heritage, we’ve just shipped a new storage offering with VFCache and announced another with Thunder, increasingly virtualization, security, management and analytics will complement that foundation to give us a broad data centre footprint.

EMC Ventures, which you run, has invested a significant amount of money into start ups of various types. It’s a lot of money. Where do you see the downstream benefit of that?

Ventures is used for strategic investments into companies which we believe are or could be critical areas to our strategy going forward. It allows us to develop ecosystem visibility which will help advance our strategy and we expect solid financial returns from those investments.

Acquisition potential?

Absolutely, some of them could be acquisitions for us. And you’re learning not to ask which ones, which is progress.

Thanks for that. Looking back at 2011 how did your team do?

We did well. There’s always room for improvement but our product leadership was seen as a key component of our great performance in 2011. We expect that to increase in 2012. For the first time ever, our mid-range products are seen as the industry leaders, a bulging trophy cabinet of industry & customer awards for that team in 2011, and we’re now competing in the low end. We’re also seen as the early leader in the big data market.

None of this was the case this time in 2011.

To you what does the competitive field look like?

In some ways, no real change from the usual. NetApp, Hitachi, Dell, HP and so on but we do see two new areas of threat in storage. One is Huawei, the other is the bevy of Flash startups. We’ve spent time looking at them both. Beyond those we see that to be the leader in the data centre market we will be competing with bigger and broader IT players in the future.

What do you consider to be the interesting trends to follow in 2012?

Cloud enabling the broad transition to IT as a service. Big Data is at a nascent stage and the impact it’s going to have on IT going forward will be dramatic. These will be what people talk about through 2012 in our industry.

You’ve run the marathons do you have a personal fitness, intellectual or artistic goal for 2012?

I’m running the Boston Marathon again with my oldest son, we’ve been talking about that a lot and I plan on doing a Century Ride with my second son. Since I need a good sit down after the training required for those I’ve been rounding out my skills studying Corporate Finance.

Being an Engineer and with all my additional technical training it was an area I never spent any time on.

Read any good books recently?

On my last vacation I read Malcolm Gladwell’s Outliers. It's a great read and I’m pondering how the ideas can be applied more directly to EMC. I also read Escape Velocity by Geoffrey Moore, I always enjoy his work.

Thanks Pat.

I’ll see you at EMC World.