The biggest challenge for enterprise IT is connecting data because unconnected data is a liability.
By the end of this post, you should be convinced of a few simple truths: first, the world we are creating is one in which enterprises with unconnected data will lose; second, they will lose to enterprises with connected data; third, they will lose because of the connected data.
Digital transformation is creating real winners and losers and the primary way to win is to connect data because unconnected data is a liability; connected data is an asset. In that world, it’s easy to see that unconnected data is a liability. Think about it this way: you cannot transform, digitally or otherwise, what you cannot manage; you cannot manage what you cannot measure; and you cannot measure what you cannot see. In the digital transformation fight, vision is life. The sighted are eating the blind and the sighted are hungry.
Let’s begin with some uncontroversial facts—stated very boldly, to increase the contrast—about the state of enterprise IT. If you don’t agree with these seven statements, then you live in a very different world than me and I won’t have much else to offer you.
Enterprises have to digitally transform. Practically speaking that means enterprises must finally learn to manage data and systems in a way that releases inherent value rather than stifles it. The status quo, that is, ad hoc direct integration between systems (what I perjoratively call “point to point” here), is how we get into this mess to begin with and it will not get us out of it.
Any asset or asset class that directly threatens revenue in the short and medium term is a liability. Isolated, unmanageable IT assets cannot be queried; their contents are unseen by reporting, apps, analytics, AI, machine learning. Unconnected data is a liability in precisely this sense: it is the key material result of the proliferation of data silos and the key obstacle to overcome to accomplish digital transformation.
Since digital transformation is a strategic goal, unconnected data is a liability because it impedes transformation. But unconnected data is a liability in another sense, too. Here I’m referring to regulatory compliance. I know, I know. Compliance is boring. Compliance isn’t sexy. I’ve heard it all before. Isn’t it just for the banks anyway? I beg to differ. You know what’s boring and unsexy? Failing to neutralize existential threats to the ongoing viability of your enterprise because you can’t see or manage all of your data.
In financial services alone, and here I’m just going from memory, we’re talking about Basel, GDPR, PSD2, and more. In other sectors we’re talking about GDPR, CCPA, and more. Regulatory compliance isn’t new. But it is growing and the global business climate suggests it will continue to grow; further, these trends are in serious tension with digital transformation efforts, with the explosion of data generally, and with the inadequacies of “point to point” IT practices. In other words, unconnected data is increasingly a literal legal liability.
In the post-GDPR world, a system you didn’t know about, containing customer data no one can remember, can get your company fined or worse. The bottom line for IT of most of these regulations is that an enterprise will be held legally responsible for data. All of its data. And it is no (legal, political, or commercial) defense to say, well, connecting data is hard; we don’t know how to do it instantly, cheaply; and, besides, it’s all Oracle’s fault anyway.
But, wait, wait, I can hear you saying, the cloud is gonna save us! Here at Very Smart Corp we are moving everything into the cloud and so ponies, rainbows, and magic.
Cloud is a big deal, obviously. It accelerates the trend of virtualizing IT assets that started with IP network technology and moved to Storage Area Networks. It precedes and rationalizes the move to virtualizing computing environments around data-center operating systems and large-scale distributed computing. The digital world is all about virtualization. Everything is either already virtualized or soon will be. Your hard drive isn’t a hard drive and your server isn’t even a computer. Old IT workers tell their children ghost stories about actual computers that, in their idle youthful days, they accidentally set on fire. And increasingly those stories don’t even make sense because what even is a computer and everyone knows clouds don’t burn, duh.
Of course data management has taken account of the cloud and embraced it wholeheartedly. Take an enterprise data warehouse, which is a pretty stale business from the point of view of revenue growth, and move that thing into the cloud and now you’ve got yourself a high flier, an equity investment rocket ship of the future! Awesome.
Except that a data warehouse in the cloud is still just a data warehouse and how does that help anyone with fast-moving semistructured data and a million office docs? This is a real problem, people, the cloud will not save you. It is here and it’s here to say, but it will not save you. Why? Because unconnected data is a liability on-prem and a liability in the cloud and, what’s worse, the hybrid nature of today’s computing environment increases the odds that data is unconnected. And, even more stark, there’s nothing about “the cloud” that makes connecting data any easier than it is on-prem. Before teasing that out, let’s step back and think a bit about what the cloud is and how we should think about its impact on enterprise IT.
We should think about the cloud like a large-scale human migration, rather than like a short-term, twitchy trend. The latter fails to take the cloud seriously enough. A migration is bigger than a trend. If you pay attention to National Geographic, you will know that large-scale human migration patterns are a fundamental driver of human history. They’re a big deal, but they are neither quick nor short. We are, for example, about 70,000 years into the original migration of the earliest humans and hominids out of Africa. The cloud is a bit like our human trip out of Africa: epochal and slow, slow, slow. When setting out on a epoch-spanning migration, you’re going to have to figure out a lot of intermediate structures and ways of life and local accommodations, long before you get to the place where you stop and settle permanently.
How is that an apt metaphor for the cloud? Because most of the work of migrating from a pre-cloud to a post-cloud world will be dominated by temporary, but still very significant, hybrid challenges of being partially there and partially not. We can concede that some day “everything will be in the cloud,” while also admitting that this day is not today, tomorrow, or any day in the foreseeable (that is, budget-implicated) future. Right now, today, and for many years to come, we will be living in a world that is a hybrid combination of some IT assets on-prem (maybe even the most strategic ones) and some in the cloud. In short, the past was on-prem; the eventual future is cloud; the present is hybrid: both on-prem and in the cloud. We already see this and it’s a contributing factor to the failure of “point to point” and it makes all the other strategic initiatives around digital transformation harder. In other words, the cloud is also a structural impediment to connecting data.
If we look closely, though, we can see an important asymmetry. On-prem and cloud are just another way of saying “unconnected data”. The asymmetry that matters here is as follows: cloud-based data management plays (and always will play) at a real disadvantage compared to on-prem data management in one crucial sense, which is all about visibility. Remember that I said you can’t transform what, ultimately, you can’t see? Cloud and on-prem can’t see the same things and that asymmetry is worth understanding. In principle, once we have moved beyond “point to point”, on-prem can see on-prem and cloud can see cloud. And even today on-prem can see cloud. But cloud cannot see on-prem and it probably never will. Don’t believe me? Well, you can see the Web from many on-prem IT assets that the Web better not be able to see. Everyone is part of “the public” and everyone can see it, but only members of my team are part of “on-prem” or have visibility into its assets. That’s the asymmetry.
That asymmetry fundamentally challenges pure-cloud attempts to unify data for two reasons: first, we are decades away from “everything in the cloud”; and, second, the only sane way to connect data that spans on-prem and cloud is to do it from systems that are themselves on-prem. No one is going to poke holes in their firewall for a cloud-based data unification service to reach deep into the enterprise and unify core strategic IT assets that were exactly too strategic (or too brittle) to move to the cloud in the first place. That’s never going to happen. And, yes, maybe some day all IT assets, strategic or not, will be “in the cloud”, and maybe some day human migrations will come to an end. Both are possibilities, but the challenge for today and tomorrow and as far as out as we can sanely plan is to find a way to create value and stable, even if “temporary”, equilibriums in the mean time. Of course pure cloud plays don’t see it this way and certainly don’t talk this way. Why would they?
Let’s summarize where we’ve landed so far. Digital transformation is hard, competitively required, and a threat. Regulatory compliance is dull, also hard, legally required, and a threat. But at least the hybrid nature of today’s IT environment makes all of this even harder. So it’s all just dark skies and rain and will we ever see the sun again?
No so fast. A fundamental law of software is that if you can’t solve a problem, then redefine the problem until you find a variant that you can solve. Does that work here? Yes. The key is to realize that digital transformation, regulatory compliance, and even hybrid computing are outgrowths or expressions of one underlying problem. Or, more accurately, there is an underlying, exacerbating cause that makes each of them worse. What if we think about all of this stuff as symptoms of the “point to point” disease?
We’ve looked so far at three big areas of IT focus. What about the other places where enterprise IT is investing? Are any of them problems that may be reducible to connected data? According to Mulesoft’s 2018 Connectivity Benchmark, most enterprises are investing in analytics, IoT, multi-cloud architecture, and AI/machine learning. Setting aside the artificiality of these categories (are analytics and AI/ML really different?), the significance is obvious: each of these areas of investment, no less than digital transformation or regulatory compliance, requires a better way of managing data and systems than the old “point to point” madness. For analytics, including AI and machine learning, data is the fuel of the whole machine. No fuel, no machine. No data, no insight. IoT is a bit different but only in the sense that it will increase the pace of data explosion, including data variety. What about multi-cloud architecture? Well that’s just another way of highlighting the special challenges raised by our age of hybrid on-prem/cloud.
In sum, most of the challenges IT must overcome going forward are either reducible to or exacerbated by unconnected data and the failure of “point to point”. If you want to bend the lines toward convergence, and be the hero of enterprise IT, you have to find some plausible system, method, and approach for connecting enterprise data, that is, reducing liabilities and increasing the value of data assets.
Here at Stardog our entire focus is building a Knowledge Graph platform for the largest global enterprises to use to connect their data, using a similar approach to the one Google, Facebook, and LinkedIn use every day to connect the web, people, and working professionals. If you recognize the challenges of unconnected data we’ve discussed here, drop us a line to learn more.
Read our Knowledge Graph 101 guide to learn more about how to connect data in your enterprise.
Download Stardog today to start your free 30-day evaluation.