Fusion-io Launched New Hyperscale Flash Product to Scale Data Centers for Enterprises, Not Just Facebook

Two years ago I wrote about Facebook’s infrastructure would be the model for infrastructure of all enterprise information technology environments.   That post was an effort to explain (at the time) why SiliconANGLE was bullish on Fusion-io when the general market was confused.  It is now what we coined as Software-led Infrastructure (SLI).  I wrote about about how SLI is a threat to Oracle.  In addition our research team Wikibon is documenting this entire evolution of SLI.

Yesterday at the Open Compute Summit, Facebook lead executive and VP of Infrastructure Jay Parikh (see picture) said the same thing.   {Editors Note: the Open Compute Project is one of single most important things happening in hardware.  It will drive what we’ve said years ago the main drivers of innovation:  big data and software-led infrastructure (SLI).  SiliconANGLE is expanding coverage in this area.  Stay tuned.}

Facebook and Apple have been the darlings of the emerging tech scene when it comes to innovation in big data and software-led infrastructure and both bellwether companies use Fusion-io technology.

Yesterday, Fusion-io announced at the Open Compute Summit in Silicon Valley their new product called Fusion ioScale. Fusion ioScale is a product that is designed for the all-flash data centers. The Fusion Scale PCIe cards scale from 410GB, doubling through 0.825, 1.65, 3.3 terrabytes.

See press release for their corporate angle:  http://www.fusionio.com/press-releases/fusion-io-delivers-the-all-flash-datacenter-for-hyperscale-and-cloud-businesses/

Fusion ioScale Benefits

Key benefits of Fusion ioScale include:

  • Up to 3.2 TB of capacity on a single half length PCIe slot, with a single controller, enabling a small form factor server to reliably scale to 12.8 TB or more, delivering maximum rack density and reducing the need for antiquated disk drive bays
  • Hyperscale servers supporting UEFI can boot from Fusion ioScale, further eliminating the need for RAID controllers or disk infrastructure in spartan webscale servers with limited space
  • Best-in-class Fusion-io endurance in all capacities, empowering caching, write-heavy databases and large scale applications
  • Enterprise reliability with the Self-Healing, Wear Management, and Predictive Monitoring capabilities of Fusion ioMemory, ensuring customers can easily meet even the highest service level requirements
  • Compatibility with the Fusion ioMemory software development kit (SDK) to leverage application programming interfaces (APIs) like Atomic Writes and directFS permitting applications to run natively on flash

Fusion-io has traction with OEM customers such as Cisco, Dell, HP Netapp, and host of other 2nd tier players.

My ANGLE – Breaking Analysis

As I mentioned two years ago in my Facebook Fusion-io post, this trend of more performance using flash is not just for Facebook and Apple but for all infrastructure environments. It’s moving fast to all enterprises and cloud providers.

Our research team at Wikibon.org has been covering this like a blanket for years. For the big players it’s been about IO centric infrastructure initially but for mainstream enterprises it’s about “Software” or software-led infrastructure (SLI).  Now it’s about setting a standard for the future.

Flash Replacing Spinning Disk (Spinning Rust)

This is an effort to accelerate the trend of moving “spinning disks” or hard drives out of the data center for good. This is a similar strategy of Pure Storage a company that’s sole mission is to kill spinning disks.

Fusion-io’s secret sauce is software and specifically software that eliminates all the traditional disk overhead. It’s not just solid state storage (which by itself is big and the new standard) but the software technology component for Fusion is game-changing which is why it’s so valuable. Fusion-io is part of the Facebook megatrend because they power the component flash and software that is a key part of the Facebook (aka cloud) “new operating system” for the lack of a better description.

Hyperscale Focus Is Where The Action Is

Hyperscale servers are for large companies or applications that need hundreds or thousands of servers to power those businesses or applications. I spoke with Facebook infrastructure leaders yesterday and they told me their product/software teams are ordering thousands of servers at a time for rollout of new features. This is hyperscale in action.

Hyperscale is already a big trend in software-led infrastructure (SLI) project.  We predict that   2013 is the year for the hyperscale conversation will move to big data.  As an example of this is Amazon Web Services (AWS) success in on boarding scale of big data projects. According to my sources, Amazon Web Service (AWS) is doing extremely well onboarding big data in the cloud.  Even though AWS is very expensive and not a perfect fit for big data, the AWS is proof positive that there is MASSIVE demand for scalable big data and software-led infrastructure (SLI) and on-premise and bare metal need to get their act together.

We see big data and software-led infrastructure (SLI) the hottest markets for Hypescale. That means the notion of DevOps and Cloud/Cloud Operation will merge with flash technology, storage, compute, software-defined networking to create this new modern infrastructure.

Modern Information Technology Infrastructure = Big Data + Software-led Infrastructure (SLI)

Competition in Flash

It’s a battle some say in the flash hyperscale market. Violin Memory Systems, Fusion-io, Pure Software, an others. I say it’s not about flash competition with each other but with the tradition spinning disk guys. There is so much market opportunity that scabbling over flash vs flash is the wrong angle.

At least at the moment there is plenty of room for all to prosper.

CEO David Flynn Talking to SiliconANGLE.tv at NetApp Event at VMworld 2012 at AT&T Park