raywoo - Fotolia
Toigo Partners International
Published: 02 Sep 2014
Talk show host David Letterman may be well on the road to retirement, but we can still credit (blame?) him for all those idiotic Top 10 articles and presentations that seem to dominate the field of technology journalism these days.
From the look of things, the solution to any IT problem can be neatly distilled into 10 convenient observations or fixes -- fewer if the publication has word count constraints.
Although such word count constraints don't exist here, I'll limit this rant to identifying the top five stupidest storage technology ideas I've witnessed in the last year or so.
No. 5: The cloud replaces storage everything
A recurring meme these days in social media and tech journalism is the "game-changing" nature of clouds, especially in terms of storage. Most recently, clouds are being mentioned to explain the dramatic reduction in year-over-year revenues experienced by the finished-array industry this year. Some pro-cloud industry pundits insist that, instead of buying more products to deploy internally, companies have been purchasing storage "as a service" from cloud providers -- an idea that, if it were true, bodes well for the cloud industry, but not so much for the storage industry.
The "shift" to cloud storage might provide a handy explanation for the confusion generated by EMC's reported 22% quarter-over-quarter revenue decline last April, IBM's 23% quarter-over-quarter miss or the falling numbers of many other storage merchants announced in Q1. The "clouds are quickening the demise of enterprise storage" narrative would actually merit all the attention pundits have paid to it, if only it wasn't so idiotic.
What really seemed to be happening in late 2013 and early 2014 was that users were hesitant to purchase new capacity before they examined their options and evaluated which purchases might best meet their needs while providing a return on their investment in the fastest or most cost-efficient way. You know, the way good IT planners are supposed to make strategic choices.
Frankly, we've been inundated by industry noise of late including competing spin around three flavors of flash storage (any of which will replace disk in some short time span), dedicated storage (designed just for a particular server hypervisor) or new (old) "server-side" (aka direct-attached) storage topologies. All of that is supposed to replace those "legacy SANs" we just deployed. No one wants to get stuck holding the bag on a new storage idea that is obsoleted by the world in less than three years.
Limited budgets have also introduced a bit of consumer savvy into the acquisition process, I'm told. Delaying purchases has the effect of putting vendor sales reps over a barrel, resulting in dramatic price reductions on storage wares. In short, a lot of storage has been sold year over year and quarter over quarter, but not at the vendor's asking price. This, more than anything else, probably accounts for deflated revenues on growing volumes of hardware.
Does this mean that clouds have had no impact? Not really. But the impact isn't coming in the form of more consumers switching from enterprise storage to cloud storage. After all, industry watchers were saying in late 2013 that cloud service providers (and not business consumers) were the main customers for cloud storage. If anything, additional demand for storage from cloud service providers should have grown the size of the storage market, not diminished it.
Oh, and let's not forget the chilling impact of the failure of Nirvanix, the No. 3 cloud storage provider (perhaps the start of a growing list for another column).
No. 4: Data reduction technologies fix everything
They don't, actually. While vendors have gone to great pains to mainstream the idea of data compression and deduplication, and to turn these functions into checklist items on primary storage, the truth is that the cost doesn't seem to map well to strategic value. Plus, when applied to deep archive, data reduction raises additional questions about the retrieve-ability of deduplicated data in, say, 10 years. No standards, no investment protection.
No. 3: Flash storage changes everything
Not really. We all like faster I/O, but the real value of flash seems to be limited to read-only data. Dynamic RAM or good old-fashioned hard disk seems to deliver better performance on writes. This realization seems to be finding its way into intelligent vendor diatribes, which increasingly cite user cases involving database index acceleration and lower latencies -- not higher performance.
Some interesting cobbles might be in the offing, including flash plus tape. Some vendors use the unfortunate moniker flape for this topology. We would advise them to check the Urban Dictionary before trademarking that handle.
No. 2: Server-side storage changes everything
If you believe that de-evolving storage -- moving from shared storage back to islands of storage -- is "progress," then this idea may have merit. In fact, server-side is at best a reflection of the loss of skills in IT brought on by the adoption of server virtualization and the inclination of IT staffs to eliminate storage specialists. Server-side is easier for a poorly trained hypervisor jockey to conceptualize than storage fabrics, with or without storage virtualization technology. However, a local storage model -- while scalable as part of a kit (think OpenStack) -- requires a lot of data mirroring between boxes to facilitate failover, vMotion and so on. Have you done the math to determine what this will do to your LAN bandwidth?
No. 1: Top [insert number here] lists explain everything
Perhaps the most idiotic conclusion of all is that list articles provide great insights. The truth is that list articles tend to provide summaries of the news that are easier for the Twitter generation to follow. At worst, they reinforce pre-existing hypotheses by supporting confirmation bias. At best, they provoke the reader to dig deeper to understand the details behind the cute synopses. Here's hoping that this list will provoke a bit of research and thoughtful debate.
About the author:
Jon William Toigo is a 30-year IT veteran, CEO and managing principal of Toigo Partners International, and chairman of the Data Management Institute.