0.011: strategies for world domination
Like many of you, I've been watching the recent discussions about who stole what from whom with more than a little bit of glee. Chris Mellor started it, Tony Pearson responded, Kirby piled on with innuendo of similar BigCo misbehavings in the past, and Storagezilla took the time to correct Tony's obvious misunderstanding of how Invista is packaged and delivered (as an integrated solution, Tony, not as a kit of bits). Oh, and then last Friday Kirby whined a bit more (guess BigCo's lawyers forgot to get a gag order, huh?).
In parallel, and seemingly unrelated to this, there have been a slew of assertions about who is really #1 in storage. Tony and IBM claimed world dominance in overall storage (if you include tape), HP claimed #1 in overall disk storage, Chris granted EMC #1 in external storage, and Kirby asserted that the real overall winner in disk storage was Seagate. Mark Lewis clarified the obvious and questioned IBM's motivations. And of course, the rest of the wanna-be's all chimed in with their own context-ridden slice of IDC's concoction of data-that-would-be-useful-if-only-it-were-verifiably-true. I must admit that NetApps' "fastest growing in multiple shrinking markets" announcement left me a bit confused, especially coming on the heels of their forecasted 6-7% downturn in revenues for calendar Q2. Given that calendar Q2 is historically stronger than Q1, I don't think these press release necessarily add up to a positive for NetApp shareholders.
This all followed by a couple of rounds of marketing 101 training between Chuck and Kirby. (Psst. Kirby - the colloquialism is "nursing at the corporate trough," even if it is occasionally pronounced "troth"!)
But today's blog entry isn't about those shenanigans.
No, I thought I'd write about what's behind all these seemingly unconnected discussions - the motivations of the leading storage suppliers that drive us to compete to be king-of-the-hill. A look into the unspoken (but quite apparent) strategies being employed to get to the top. And my anarchist's view of what they/we are all missing.
Ready? Read on...
intergalactic business machines
Make no mistake, BigCo is big. And they intend to get bigger - shareholder reward is intimately coupled with growth.
Even if they haven't outright admitted it, their strategy in the storage space is one of commoditization. After several apparently failed attempts to actually compete in the storage market, IBM today OEMs their entire mid-tier disk storage platform lineup. At the high-end, their so-called enterprise products are built around clusters of general-purpose servers running I/O translation and copy functionality on top of a general-purpose operating system. And their SAN Volume Controller is different I/O translation and copy software running on a different general-purpose operating system running on essentially commodity PC server hardware platform.
The Objective? Commoditize the value-add in storage to the point of extinction, choking off the revenue stream of competitors who otherwise just limit BigCo's growth. Sooner or later, there'll be no more direct competition, and storage can be profitable once again. Worked for the mainframe.
The Tactics? Replace, displace and erase. Create barely credible products out of commodity parts, market them as the greatest thing since the transistor, and force them upon every customer you can - heck, bundle them for free with new mainframes and mid-tier servers if you have to. Dumb down any unmovable competitor storage by front-ending it with low-cost, commodity components with "good enough" departmental-scale features that neuters any competitive value. Shrink the overall storage market by relocating the revenues into Services wherever possible. Invest in data reduction technology not just because it's "Green," but to further reduce demand for competitor's premier storage products. Restrict access to host platforms through licensing & electronic signatures. And OEM any competitors products you can, especially in the fast-growing mid-tier, and then choke off their revenue streams through discounts, concessions and patent threats.
The Weapons? No-cost bundling, host & application integration with proprietary storage services, delayed and expensive licensing of new host & storage APIs. But most importantly, nobody can compete with IBM on commoditization of technology, if only because IBM's installed base is sufficiently large to drive huge demand quantities. And credit to the current management, who are demonstrating an uncanny skill at getting OUT of a market just before commoditization destroys the ROI. Oh, and don't forget the power of Global Services - even if you have to operate at a loss until you can dictate the next technology refresh supplier.
I call this the "Attack of the Clones" strategy.
universal storage platform
Unlike IBM, Hitachi doesn't even try to hide their strategy - it's been emblazoned right there on the front of every one of their high-end arrays since 2004. And it doesn't stop there, with software titles like Universal Replicater and Universal Volume Manager. Heck, they even insist on calling their storage array a "Controller" in what could conceivably be more than just an assertion of control, but also of dominance over the SAN "Director" suppliers.
The Objective? Absorb, neutralize and eventually displace the competition by installing proprietary volume management and I/O redirection in the storage I/O path.
The Tactics? Sell anywhere and everywhere the competition isn't to create a perception of broad market acceptance. Avoid complex implementations where ever possible to avoid cost and risk. Actively market directly to the Wall St. analysts to gain mind share and undermine the competition's market cap. Pronounce an endless stream of exaggerated specifications to create the impression of technical superiority. Leverage global OEMs and resellers for market coverage, but don't hesitate to take the business direct or to undercut the channel if necessary to drive revenue and footprint, especially at the end of the fiscal year, or if the deal is on Wall St. Penetrate competitive strongholds with the lure of below-cost storage to get a Trojan Horse footprint for the future, then increase pricing once securely integrated into the IT environment. Under the guise of benevolent unification, position in-band I/O virtualization of competitive products as a means to extend their life. Force virtualization on any customer who wants to tier their storage by not supporting low-cost storage alternatives within the Tier 1 storage array. Leverage this position to neutralize any competitive functional advantage (sound familiar?). When the inevitable performance, availability or cost-of-maintenance issue arises, blame the competitor's storage and move the data seamlessly into the array where it can be safe and controlled.
The Weapons? Strong tier 1 technology backed by availability guarantees. A compelling vision of centralized control and demonstrable single-system "ease-of." Endless capital reserves of an off-shore conglomerate and an arms-length relationship with a marketing subsidiary that doesn't have to comply with the same business reporting rules as the US-based competition. And a strategic relationship with BigCo, born out of a negotiated exit from the mainframe market, that unifies the two companies against the common enemy (this weapon even has a name: ABE - Anything But EMC).
I've come to refer to this as Hitachi's "Trojan Borg" strategy...
evil machine corporation
Then there's my employer - the storage vendor everyone loves to hate. And not without reason - when you define and create multiple markets and then execute well enough to establish and maintain the lead, it's inevitable that your competition makes you out the bad guy.
The Objective? To be the preferred supplier of information infrastructure, by offering technology solutions that address an ever-widening spectrum of information retention, protection and distribution throughout its lifecycle.
The Tactics? Listen to the needs and challenges of high-end customers, and address them before they have reason to seek alternatives. Leverage every technology possible to reduce costs and expand the portfolio to cover the widening information infrastructure footprint. Business continuity and disaster recovery are the cornerstone of long-term success and stickiness in the enterprise space - deploy these everywhere. Think Big, even when addressing small or emerging markets. Failure is not an option, #2 is never good enough, but arrogance is an avoidable derailer. One size doesn't fit all, and it is preferable to offer overlapping solutions rather than to have gaps between product lines. When customers want one throat to choke (or trust), be there. Focus on more than just the storage itself - as the hardware (inevitably) commoditizes, derive revenues from an ever-increasing stable of storage, information and infrastructure management software. Security isn't a bolt-on; managing complex information *and application) networks is too complex to do by hand; And if we can't build it fast enough, buy it.
The Weapons? Best-of-breed technology and market share leadership in virtually every measurable market we're in. An absolute commitment to the customer satisfaction and retention, backed by a global buck-stops-here service organization. Unmatched investment in interoperability qualification. More than 16 years of experience protecting the information assets of the world's most demanding enterprises, and the trusted advisor expertise to leverage this for enterprises of any size. Products designed to scale up, out and down, architected to optimize TCO for the customer, and implemented to minimize operational complexity at virtually any scale. Great relationships with key partners in specific segments to complement a global direct sales force that is second to none. Strong capital reserves and fiscal responsibility. And a diverse, ambitious and accessible management team.
EMC calls this "Information Infrastructure"...but I'll call it "The Big Foot(print)" strategy.
but nobody is winning!
OK, I apologize that this assessment is biased and one-sided. Unavoidable.
But the important observation is that these strategies aren't being successful. At least not yet - so far there is no clear winner in the storage space. Sure, IDC data says that EMC is #1 in External Storage, and by our own internal estimates we're #1 in both the high-end and the mid-tier markets. But since the market rationalization of the early 2000's, there hasn't been a clear break-away from the pack. And honestly speaking, there doesn't appear to be one coming any time soon.
I think I know why: None of us have effectively solved customers' biggest storage problem: scale.
And not just the problems of exa-scale customers that I discussed last week - I'm talking about the scalability requirements of virtually every IT infrastructure there is, from Google to Ed's Pizza and from World Bank to the storage network in your own home.
Sure, we all have visions of how to scale. Some predict an eventual "information grid" with unlimited access to information (and information storage) available from a wall jack, with SaaS and SOA and Xyz 2.0 as the dial-tone.
But what are we going to build this out of? One vendor thinks it could be with lots of little nodes in front of our storage. Another proposes a centralized intelligence redirecting I/O to the appropriate location. And the Evil one proposes purpose-optimized storage with an intelligent infrastructure.
None of these approaches scale adequately today to meet the demands of the exa-prises. And even if they double in scale every 2-3 years (the apparent product development cycles for major leaps in the storage industry), they can't keep pace with 56% CAGR growth of information. Fact is there are more IT organizations every day struggling under the burden of information growth, flat-to-shrinking budgets, and inflexible technologies. The exa-prises of today already require multiple DMX-3's or USP-V's to consolidate their storage, while the newest breed of storage consumers (e.g., Google, Yahoo, YouTube, MySpace, Second Life, etc.) have already surpassed any practical application of such "archaic" tier-1 storage solutions. Looking forward, it will likely be the rare customer (or consumer) whose information infrastructure can safely and sanely be deployed on a single box or with a single centralized intelligence.
the anarchists own
I do have a plan. Or perhaps a strategy. OK, maybe it is just a vision. And although by definition EMC owns it, it is in fact my own. I'm not aware that EMC has either conceived it or specifically actioned it. While I see some things we're doing that could help, I can't promise we'll ever deliver upon my vision, but I do think it's the only way to solve the "scale" problem. Don't forget my disclaimer:
The opinions expressed here are my personal opinions. Content published here is not read or approved in advance by EMC and does not necessarily reflect the views and opinions of EMC.
From my perspective, what's missing in all these strategies is the ability for all the different solutions to pro-actively cooperate and integrate. In fact, it seems inherent to each strategy to limit and even prohibit cross-platform integration. Sure, we have basic protocol interoperability, but those protocols are old and they were never meant to deal with the challenges of scale that we face today. And we have standards bodies trying to build management APIs and standards, but vendor interests have nullified most of these to the barest of common denominators while vendors are pushing customer implementations toward their not-so-common proprietary features. We even have EMC ControlCenter (what's in a name?), which does a pretty good job at scaling, but still has a long, long way to go.
The global telephone systems work because of standards established in the early days of telephony, and most of this natural interoperability has expanded to encompass cellular wireless communications as well. Similarly you can today take practically any electronic appliance to any country and power it with little more than a couple bucks worth of incremental expense for an adapter.
Not so with storage. Choosing a solution almost always choosing a vendor, and accepting that vendors' strategy for world domination.
why can't we all just get along?
Actually, my vision is rooted in the concept of anarchy - specifically, definition #3 at dictionary.com:
anarchy (noun): a theory that regards the absence of all direct or coercive government as a political ideal and that proposes the cooperative and voluntary association of individuals and groups as the principal mode of organized society
Imagine: storage products that inherently unite and work together to provide your enterprise with storage services. Not necessarily total anarchy, but inherently cooperative in support of the customers' best interests instead of the vendors.
In my vision of anarchic storage, things work something like this:
As new storage products and technologies are introduced into the infrastructure, its capabilities are inherently recognized and utilized to "improve" the infrastructure. Data is stored in accordance with its requirements for performance, availability, protection, and value. If a storage device that can power down disks is added, the appropriate data is transparently relocated there to save power and cooling. A device with fast storage (e.g. solid-state disk) is automatically utilized for response-time critical data, or a caching node of DRAM might be autonomically employed as a high-performance buffer. Backups and remote replicas are automatically created and retrieved on demand in the event of a disaster or accidentally deleted/corrupted file. Data objects that are being maintained for compliance are immutably protected and retained on the most cost-effective media possible.
Not that the storage administrators job would go away, but the job title would probably change to "storage advisor." The job would be more about describing policies for the data, and perhaps responding to the storage infrastructures requests ("need more cache to meet the performance SLAs" or "replace node XX because the power/failure rate is too high"). In the beginning, there'd be lots of monitors and dashboards to make sure that the infrastructure doesn't pull a HAL on you ("I'm sorry Dave, but I can't do that!"). But over time, as trust grows in the environment (and the vendors who supply the components), the storage anarchy would require less and less actual monitoring - much like the power grid and telephone networks of today.
The key enabling technology for this isn't really all that hard. In fact, it probably requires more social engineering than technology...vendors would have to learn to make money WITHOUT forcing customers to make lock-in choices. Oh, there'd still be lots of room for differentiation - potentially even more than today, because start-ups would able to more readily integrate their innovations into virtually any data center, instead of being locked-out by the dominant vendors. At the core would probably be the standard for storage self-description - kind of like the USB protocol, which allows any USB device to explain its capabilities to whatever it gets plugged in to. Storage anarchy would also require a complimentary protocol for applications to explain the services that they require. SOA and SaaS touch on this, but don't go far enough in my opinion - we'll need a lot of descriptors about how the applications not only use the data, but how they expect it to be protected, when/how often they need the data, and any interdependencies. Even the server and network components will have to be able to describe their capabilities and their requirements to the infrastructure.
But standards alone won't drive or enable the solution. We'll also need an embodiment of the anarchical decision maker mechanism / organism.
At the core of my vision is a de-centralized, distributed and unselfish intelligence that marries storage capabilities with application requirements and the overriding IT policies. Decentralized and distributed because that's the only way to scale this whole idea up - done right, a companies IT infrastructure should be able to span the entire globe, with local optimization within a global context. For some enterprises, the infrastructure will even have to reach up to non-terrestrial information resources, be they on the Space Station, on the Moon, on Mars or on the way to somewhere else.
Most importantly, this intelligence will have to be even-handed and unbiased, no matter how or where it is implemented. With common guiding principle of "make things better" (where "better" is defined by the IT Policy), old storage will inherently admit when it is old and ready to be replaced, and new storage won't undertake responsibility for things it can't handle. Servers, Operating Systems, Databases and even Applications will cooperate for the greater good. For example, File Systems would tell storage when a file has been deleted so that the space can be reclaimed, or File Servers will concede data to a faster device when they can no longer meet the target SLAs. And all this would happen transparently, with virtually no human intervention. Just one, big cooperative storage infrastructure, humming along virtually unnoticed (at least as long as the objectives are being met) - just like the Power and Telephone infrastructures (ideally without any of the outages, though).
So, to summarize:
The Objective? A level playing field that allows customers to truly choose and deploy best-of-breed storage infrastructure technology solutions without any form of vendor lock-in.
The Tactics? More vendor time/money/effort spent on being best and first, and less on trying to be everything to everyone. Innovation gets to market faster, and can support the needs of a larger base of customers, no matter how small the inventor. No need for three different and incompatible replication solutions - the focus would be on improving implementations of the first one. With fewer competing alternatives, complex tasks such as interoperability testing eventually simplifies to the transparency of turning on your cell phone. And the "management" of the infrastructure evolves into near non-existence, replaced by policies that become the guiding principles of each individual IT shop's information infrastructure - the rules that allow Anarchy to rule!
The Weapons? Standards, models and simulations will be the string that ties this all together. But ultimately, customer IT budgets will have to drive the change, either by forcing today's vendors to play by new rules, or by choosing next-generation suppliers who have already adopted a notion of communal success through anarchist cooperation.
Oh, I call my strategy "Storage Anarchy" and the unifying intelligence "TWINE"...
(I'll bet you thought I'd call it The Matrix ).