John Glendenning examines the rise of all-flash/HCI mania, and explains why a correctly-provisioned and cost-effective approach is a better fit for many businesses.
Have you developed ‘must have all-flash’ or ‘shiny new HCI toy’ syndrome?
If you have – don’t panic. You’re not alone as many share your affliction and it’s probably not too late to take a step back before you spend your organization’s precious cash.
I spend a lot of time with organizations, both end users and their integrators, and two concerning trends are becoming common. The first is ‘have-to-have all-flash syndrome’ and the second is ’hyperconverged new-shiny-toy syndrome’. These afflictions often hit at the same time with an all-flash HCI solution being deployed as the sledgehammer to crack the metaphorical walnut.
The overall trend is primarily characterised by the purchase of a solution that is significantly over-provisioned for the target workload, both now and in the future. This is happening all the time with solutions being deployed with capability as much as 10x greater than required now or in the future.
Here is a prime example of over-provisioning at the far right of the scale: StorMagic exhibited at VMworld this year and during the show an unsuspecting IT professional approached the booth to find out what we were about. It quickly became apparent he had recently deployed an all-flash HCI solution from Nutanix. My colleague assumed he had pretty intensive compute and IOPs requirement and was surprised to learn the company had a handful of small sites with no more than 50 people at each site and your basic back-office apps, file, print and the usual network services.
It turns out that they had a single server solution previously and wanted redundancy. They were sold on HCI and did what his integrator told him. We estimate his 3 server all-flash Nutanix solution was 10-20x overspecified with an IOPs potential of 5-10k sold against (at most) an IOPs requirement of 500-1000.
We demonstrated to the latest victim of these syndromes that a StorMagic powered off-the-shelf server solution would have more than met the requirements at a cost of $10-$15k and not the $40k he spent per site.
Meeting requirements and saving money
Here’s another example of a StorMagic customer that we managed to engage early in the buying process and before they had made the wrong decision. A national brand retailer that was just about to deploy all-flash arrays in all their stores. We demonstrated through detailed IO analytics that magnetic drives with memory caching would more than meet their requirements and saved them a cool $14m on their server configurations.
Okay, I get it; perhaps this is starting to sound like a StorMagic advert, though I make no apology. We come across these examples all the time and I think the reason for this is our focus on the edge of the enterprise. Often the same IT professionals architecting the datacenter are architecting the infrastructure for the edge – and old habits die hard. With our focus on the edge we likely bump into examples of over provisioning more than most as these datacenter architectures get squeezed into the needs of the edge of their IT operations.
If you’d like to find out more about how StorMagic helps businesses around the world achieve their storage goals, click here. If you’re ready to get started today, why not check out our free 10 day SvSAN trial?