I’ve been imposing on the world sharing my theory on the formation of an information utility for some time now. Various big vendors have announced alliances with that kind of aim.
For instance, Oracle and Microsoft earlier this year announced an effort to make their cloud databases interoperate, and Microsoft and others have announced common data model projects.
Now Salesforce, AWS and Genesys have teamed up to announce their own common data model project with The Linux Foundation. The effort purports to be an open source common information model, or CIM.
This is a good idea in some respects but it falls short in others.
We’ve been living in an era of open standards since at least the 1970s, when standards like SQL became viable and the industry pivoted — deciding that relational databases were far better than the flat file systems that then dominated IT.
For a great explanation of how design and standards power the tech world, you can’t do better than Baldwin and Clark’s 2000 Design Rules from MIT Press, and note the double entendre on “rules.”
Building a set of common standards is perhaps more important than an invention itself, Baldwin and Clark essentially argue, because the rules let any number of players contribute products, services and knowledge to a particular paradigm.
The opposite of open standards is a siloed world where nobody talks to anybody. Imagine if tires were specific to a particular brand of car, and you’ll realize what an impediment to innovation a lack of standards poses.
This is why I’ve been so up front about watching an information utility form. As I see it, IT is so pervasive, and it is commoditizing so quickly, that the old structures we once used as differentiators are now inhibitors to further progress.
Whose database you use is practically irrelevant today, except if you need high performance and resilience, in which case I can’t think of a better choice than Oracle’s Autonomous DB.
Oracle may be the only thing standing in the way of totally commoditized databases, but that’s a digression. Clearly, a part of Oracle’s success is its hardware, and it would be wise for the company to offer its Exadata technology as a kind of standard for competitors to use too.
Common Information Model
So Salesforce et al. want to promote an open standard, but there are issues to consider. A standard isn’t a standard in any real sense unless it is pervasive and has knocked competing “standards” out of the competition. That hasn’t happened yet, though it’s a logical consequence, and I have no idea which standard will prevail at this point. I suspect that none will. I foresee a few similar standards coexisting.
The proposed cloud standard can be less dogmatic than SQL, for instance, and I expect that the arriving IT utility will employ a few standards that, as I have said before, will give the appearance of a homogenized utility.
Under the surface, though, there will be ample opportunity for vendors to differentiate. My favorite example is the electric grid, which is really an assembly of niche providers adhering to standards that give the appearance of a national grid — but nothing could be further from the truth.
IT and data don’t fit neatly into a cubbyhole with one correct way to do things. There are lots of ways to build apps that do pretty much the same thing, even though they may capture and store data in vastly different ways. A common data model’s importance is its ability to set a standard so that all people who work with it have an understanding of how they differ, but also how they integrate.
So, the talk about common models and utilities really boils down to how things efficiently integrate, and The Linux Foundation announcement says as much:”The CIM reduces the complexities of integrating data across cloud applications by providing standardized data interoperability guidelines to connect point-of-sale systems, digital marketing platforms, contact centers, CRM systems and more.”This places significant reliance on any vendor’s integration functionality and proves the value of Salesforce’s US$6.5 billion acquisition of MuleSoft in 2018. MuleSoft and other high-end integration services don’t start with writing custom code to enable point-to-point communications between apps. They use APIs that make every app a bidirectional data gateway to the world.
Having a common data model that all integration tools can make APIs for is really the power of this approach. It’s also the glue that will hold the information utility together.
My Two Bits
This leaves us with the reality of competing standards for data and integration. As the various standards take hold, we’ll no doubt see great progress within discrete alliances for making their apps all play nice in the sandbox. Soon enough, though, we’re going to be asking the same questions about interoperability we always ask when we realize we’re just in a bigger silo.
This in no way argues against making the attempts we see in the market today. We’re simply seeing the evolution of new standards and perhaps an information utility.
However, if you are a vendor and not named Oracle, Salesforce or Microsoft, for example, it would behoove you to take a hard look at your code, your alliances and your markets, and ask the basic question: How will you get along in this landscape five years from now?