Every now and then I run across something in a blog that is so aligned with what we are thinking, it feels like I couldn’t have written it better myself. It happened two weeks ago, when no sooner had we finished our session with Steve Herrod about the future of storage (“Is Storage Dead?”) that I read a post by storage guru, and aspiring rock guitarist, Steve Duplessie, who seemed to be reciting from a very similar hymnal.
For those of you that missed it, Steve Herrod (managing partner at General Catalyst Partners and former VMware chief technology officer) and I recently had a lively webinar chat and Q&A session where we decided that the storage industry is alive and well, but that the current model of acquisition – i.e. buying and caring for boxes – is most certainly on the road to oblivion. It is gratifying to read that Steve Duplessie is seeing the same thing from his end of the data universe. His money quote is below:
“When we ran out of something, we bought another box. And another. We loved our first box, and hated our 87th. Such has been the life of the infrastructure buyer.
Today we don’t (or shouldn’t) buy like that any longer. Those metrics don’t matter anymore.”
Scaling, cloud strategies are limited by storage boxes
Having spent the last three years studying what is ailing the enterprise, we have found that a proverbial 87th box is a source of much consternation. IT managers who have only one or two arrays don’t often understand the problem of scale that persists in the enterprise. At some magnitude, it simply is infeasible to have only the fastest flash everywhere. So instead, storage managers build silos of fast, medium and slow storage. They try and guess which workload belongs in each silo. They protect against disasters, hardware failures, and human errors through redundancy. Then, they try to scale each silo so they don’t blow the budget, while their vendors demand forklift upgrades every few years.
This isn’t a problem, as some suggest, that is solvable with software. Adding more software to this teetering pile of complexity can only make things worse. Breaking it up into even smaller boxes and claiming it’s “converged” isn’t the right answer, either. IT wants all of these boxes to go away. Companies want a cloud model. They want to be freed up to deploy applications for their users without having to worry about data placement, cost, and, even worse, running out of space. Having to truck in a new box whenever you need to deploy an app isn’t exactly agile in the post-virtualization age.
Freedom from the storage treadmill is critical for TCO
When I talk to analysts, we tend to dive into the details of all of the innovations we have built into our global storage network – enablers for new use cases around data mobility, disaster recovery, and hybrid cloud computing – but our core mission has always been to take away the boxes. The never-ending treadmill of building out and tearing down row upon row of infrastructure was always the most intractable problem customers faced in my past life at EqualLogic and Dell. This was what motivated the idea behind ClearSky Data. When we talk to customers today, it continues to be the issue about which they engage the most. When we discuss total cost of ownership (TCO), we find that removing the cost of this treadmill is the most compelling argument for a cloud business model.
Hence, this is why we built our global storage network – an entire technology stack built around the requirement that we can deliver storage as a service across a network. We needed to leverage the management, scale, and economics of the cloud. At the same time, we need to deliver enterprise-level performance, availability, and features, as well as best-in-class performance and cost optimization. It’s great to see that smart folks including Steve Herrod and Steve Duplessie are starting to envision a future without storage boxes cluttering its path too.
Avoid box storage overheads while meeting the performance and latency needs of primary infrastructure.