How to overcome the infamous forklift upgrade challenge

shutterstock 306988127

Thinking about modernizing your IT infrastructure? If so, it’s likely you face two major pain points: moving to a new modern storage infrastructure and keeping that new infrastructure current.

Some vendors categorize these pain points as the “forklift upgrade challenge.”For those not familiar with this term, a forklift upgrade is that portion of the storage lifecycle when someone drives in a forklift to haul out the old hardware. Unfortunately, most vendors focus only on simplifying the storage lifecycle after dismissing the legacy storage solution. They have little to say or offer about how they can help you migrate from your old storage infrastructure. Some storage vendors approach the problem by offering the ability to replace controllers to deliver more IOPS and support future software features, adding different generations of equipment to an existing cluster. At first, this looks absolutely fantastic. Solid state drives (SSDs) and flash-based media are not susceptible to the same mechanical failures common to hard disk drives (HDDs). This enables you to use your SSD asset longer and further maximize your storage investment, but is longevity of the NAND flash alone good enough?

Longevity and the bathtub curve

All mechanical and electronic components tend to follow the bathtub curve. The curve is widely used in reliability engineering and its name is derived from the cross-sectional shape of a bathtub: steep sides and a flat bottom. In a nutshell, while NAND cells indeed have a lower failure rate, other components of the SSD—like the drive controller and other system components, including drive enclosures, SAS connectors, etc.—follow this curve. It’s interesting to note the findings from a recent study that highlight Google Inc.’s experience with SSD reliability in the real world. The study discusses how an SSD’s age, not usage, affect its reliability. You can read the complete report from Google and the University of Toronto here.

Only as fast as the weakest link

Another downside of those approaches is that the overall system will only be as fast as its weakest link. To use a car analogy: I can replace the engine on my old car with a more powerful one to improve its performance, but if the tires and transmission are not also replaced, then a new engine is going to be of little use. Even if I replace the tires, there’s only so much power I can put through the car’s old drivetrain, a fundamental part of the original design, before it will literally tear itself to pieces. The bottom line is this: A car, just like a storage system, is only as fast as its weakest link. So replacing only a given component and/or mixing different generations of hardware might not necessarily produce the desired results.

Making room for storage innovation

Getting all of the above to work (an intermix of different generations of hardware) comes at the expense of innovation as the next generation platforms must rely on obsolete hardware architectural decisions. How can you leverage the latest PCIe, SAS, or Intel NTB (non-transparent bridge) technology if the new generations aren’t backwards compatible? As a matter of fact, many standards take a long time to be ratified. Eventually, when the lifecycle of the entire system comes to an end, how do you migrate the data? Does it make economic sense to pay for maintenance and support costs on the backend of the array with a new frontend, especially once the depreciation cycles are over?

Take a very different approach: the 3PAR federation

We have looked at the heart of this challenge, the act of migrating the data, and come up with a solution that enables you to move to 3PAR off a third-party array and streamline the entire 3PAR asset lifecycle management and technology refresh. We call our solution HPE 3PAR federation. HPE 3PAR federation technologies elevate data to newer hardware technologies while avoiding the pain associated with forklift upgrades that require extensive planning and preparation (and may include such difficulties as application downtimes or other impactful events). Unlike other technologies, 3PAR federation gives the option to either repurpose legacy arrays, extend their value through further federation or replication, or simply retire them. Most importantly, because HPE 3PAR storage federation is an enterprise solution, it won’t present a  single point of failure (like running on a single controller). It also supports online migration of volumes that are exported to clusters and/or replicated. This approach solves your refresh-triggered data migration pain point without introducing risk and without compromising on hardware innovation. The HPE 3PAR architecture and flexible software stack deliver continuous software features throughout the lifecycle of the product.

Our software releases over the past three years are testimony to innovation and our commitment to existing customers. As demonstrated by our SPC-1 and SPC-2 results, the HPE 3PAR architecture is the most efficient in the industry when it comes to delivering performance. This means that you can count on having enough horsepower for years to come.

Want to know more?

Download and read the Storage Federation white paper here.

Ivan Innaconne is Worldwide Product Line Manager for HPE 3PAR StoreServe Storage at Hewlett Packard Enterprise in Palo Alto, Calif. He is a senior IT professional, who brings more than 17 years of academic and professional work experience in IT-related fields to his current role at HPE.  As Worldwide Product Line Manager, he is responsible for developing the next generation product roadmap and product strategy for 3PAR. Innaconne, a native of Italy, is an avid wine enthusiast.

The march toward exascale computers
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies