At Oracle OpenWorld, Fusion-io announced the next evolution of its ioMemory platform for application acceleration and data center optimization.
According to Fusion-io, its storage memory platform for shared data decentralization improves the processing capabilities within a data center by relocating process-critical, or "active," data from centralized storage to the server where it is being processed, a methodology referred to as data decentralization. Fusion's data decentralization platform allows enterprises to consolidate or reduce complex and expensive high performance storage, high performance networking and memory-rich servers.
The company's first generation platform was brought to market 4 years ago and it has largely been unchanged but because it is a very software-centric platform, the company has been able to more than double the performance of that product and go from 5X nanometer NAND flash geometries through the 4X and to the 3X geometry NAND flash that is shipping on the platform today. "That has allowed us to double the capacity point. With software enhancements, we have been able to double the endurance and we have been able to extend the reliability of the platform, all within the context of that one hardware platform, which is today the gold standard when it comes to application acceleration capability," David Flynn, Fusion-io CEO and chairman, tells 5 Minute Briefing.
"With the introduction of this next platform, we are stepping down to 2X nanometer NAND flash which is doubling the capacity point," says Flynn, explaining that as the capacity point is expanded, opportunities are opened up for it to be used in new areas. "People want to get the most capacity that they can because it enables them to do things that they couldn't otherwise do." In addition, he notes, this capacity is at a much lower cost point, so much of it is coming for free because of the cost reduction of going to higher density NAND flash.
"The value that we sell isn't a storage device with some speeds and feeds and capabilities. What we sell is the ability for your server to do 10X the workload - and you only have to pay that software licensing once, and you are only maintaining one server. One way to think of it is as a utility multiplier. By improving the output per server you are improving the effective cost per unit output proportionally of everything," says Flynn.
Specific new features of the ioDrive2 and ioDrive2 Duo include:
- Nearly symmetrical read and write access with low queue depth performance
- Consistent low latency access for mixed workloads with 15 microsecond write latency, 3 GB/s bandwidth, over 700,000 read IOPS and over 900,000 write IOPS
- Maximum capacity in the smallest footprint, with up to 2.4 TB of capacity to fuel server efficiency even in the most data-intensive environments
- A new intelligent self-healing feature called Adaptive FlashBack that provides complete chip level fault tolerance, enabling ioMemory to repair itself after a single chip or a multi-chip failure without interrupting business continuity
- System level integration with the latest Fusion-io VSL 3.0 software subsystem
- More work per unit of processing with the cut-through Fusion-io architecture, which continues to deliver performance increases as CPUs become more powerful
- Extended support for all major operating systems, including Windows, Linux, OSX, Solaris x86, ESXi 5.0 and HP-UX
- Field programmability, allowing IT professionals maximum flexibility in customizing their ioMemory platform to meet the unique needs of their enterprise
"We live in the information age. The whole point of the information age is to derive value from raw data by processing it into something that is more useful, kind of like manufacturing takes raw material and turns it into a usable good. The raw material here is data and the usable good is some form of knowledge that gives you value. The better you can deliver the data, the more efficient you can make those factories," Flynn explains.
"Today's workers - the server and the processors - are really very idle and the reason is that you can't get the material into the assembly line fast enough - and that is because RAM is too small to hold much and disk is too slow to deliver it very fast. So while one has lots of capacity it doesn't have much performance and while the other one has lots of performance it doesn't have much capacity," he says, noting that Flash represents the perfect middle ground.
According to Flynn, Fusion-io has over 1,500 customers that span enterprises large and small as well as web properties. Customers include craigslist, Zappos, Answers.com, and Facebook, and the applications for Fusion-io are as varied and diverse as the applications of RAM and of disk.
"We are excited at being at the forefront of this technology space because it gives us a chance to work with customers who are pushing the envelope and trying to do things that would not have been cost-effective before," says Flynn. "Every major business today gets a significant amount of its value from how fast it can process data."
The Fusion-io MLC ioDrive2 and ioDrive2 Duo will be offered in 365 GB, 785 GB, 1205 GB and 2.4 TB capacities, beginning in late November 2011, followed by SLC products in 400 GB, 600 GB and 1.2 TB capacities, with the first product run already committed to key customers and partners.
To learn more about Fusion-io, go to www.fusionio.com.