The MDM Silo Problem Vendors Are Ignoring

 Aaron Zornes (@azornes), chief research officer for the MDM Institute, says large companies now have multiple MDM deployments, and need a way to manage these siloed solutions. The vendors tend to deal with this with solutions that only address their products. A more practical solution for most companies would be a registry-style MDM hub, Zornes told IT Business Edge’s Loraine Lawson, but so far, that’s an option vendors have ignored.

Lawson: So we have multiple MDM tools coming into the enterprise. Do you have an answer to handling this problem?

Zornes: We don’t have an answer. Some vendors, like Collibra, which is, you know, a 10-person company out of Benelux, that’s interesting stuff. They're basically a bunch of university PhD types who have a good product that is linked to IBM and that Cognizant is using and Orchestra Networks is working with, I think, and a couple of other people. But they're a very small company.

It’d be smart if Informatica bought Collibra or somebody like Informatica would take one of these uber-type hubs, or hub of hubs, whatever you want to call it, and make that an offering that would re-unify, re-patriate the data and the different customer hubs that you have and so on.

So that’s part of the dream that’s going on now. It’s how do we get to that stage? That’s where another interesting technology comes into play called Registry. Do you remember registry servers or registry hubs for MDM at all?

Lawson: I do.

Zornes: The prologue is the MDM vendors would like us to have collective amnesia and forget about registry hubs. But Initiate systems, Microsoft, Siperian, Sun, Hyperion, H-Y-P, Oracle, these were all registry hubs, registry MDM solutions, registry architecture.

A registry is just an index that carries keys pointing to other databases. So that’s all Initiate was and that’s what Siperian was initially and that’s what Sun Enterprise MDM was and that’s what Hyperian MDM (note: Now Hyperion DMR) is. And that’s what Microsoft’s Master Data Services was. Informatica bought one too, called Identity Systems.

What a registry hub would do is it would say, “Well, you know what, you don’t have to move all your data into one database. That’s too hard. That’s too complex. You’d have to rewrite all of your applications. It’s politically incorrect or impossible or physically impossible.”

So what they're going to do is create an index layer over your master data that will aggregate it, materialize it on the fly. Composite Software talks about virtual MDM views. So that’s what Initiate and these other characters did.

If you're a health care company that’s got all these fractured health care systems that all have Social Security as their master ID, they're not supposed to be using it, at least in the U.S.  So you roll into a health care company with Initiate Systems and you say, “I’m going to install the software. It’s going to go through all your databases one time and match everything and clean it up and we’re going to create a super index, like an index to a book or a portal that just looks at all your data. So you can type in the Social, the account number and it will go out to all the different databases, and bring all the data that you ask it to bring together for that one customer view.”

But it will do it on the fly. It doesn’t persist it, okay? So the only persistent data is probably name and address, maybe phone numbers, and keys to other systems.

They could come in and, in a day, half day or two weeks, have you up and running. If you brought in IBM, Oracle, SAP, a year and a half later you're still struggling to get the value to get the system up and running because they're a different type of architecture. The IBM, Oracle, SAP systems said, “We’re going to determine what data is going to be mastered and we’re going to persist it in a central database and then we’re going to replicate it back out as needed or as requested in the published subscribe model, but basically then we have to rewrite the applications to no longer look at their old master data source, but to look at the new master data source because that’s where the master data now is.”

That’s the real view that everybody wants, because that’s better architecture, long-term and performance-wise. But meanwhile, Initiate caused IBM a lot of pain. DRM (Data Relationship Management) caused people pain, because they could come in, in a matter of hours or days, to do what it took the other guys months to do. But that’s all they could do. They couldn’t do all the other heavy lifting without getting into the details.

So that’s the registry solution.

Informatica bought one of those called Identity Systems from Nokia. IBM bought Initiate Systems. Oracle bought Hyperion and ended up with one. Microsoft bought Stratature. So they all had these index systems that could very quickly create a view of X, but couldn’t take you into the broader, deeper MDM capabilities.

Then the challenge is now that I’m IBM, I’ve got the persistent hub that was originally DWL, which became Webs for Customer Center, which became MDM server. So IBM MDM server is the old DWL Webs for Customer Center product, which was a very fast high-end system that banks, retailers, phone companies, insurance companies all liked to use, because it does a really good job of mastering MDM.

But a lot of people are being sold this Initiate version. The question is how do I get from the Initiate Systems version into the full-blown MDM hub? And that’s where there’s a little bit of sleight of hand in the marketing going on, because it’s not an easy—it’s not just flip a switch and you're there.

It’s the same with Informatica, the same with Oracle with DRM and their MDM systems and so on and so on. These registry systems are seductively good, because they provide value, they solve problems for us, but they're not the real MDM.

But, what we’re now learning is, you know what? Those registry systems could solve this problem for us of how do we manage data across multiple MDM hubs. Why can’t we create a registry that spans different MDM hubs? Why can’t we create a hub that looks across all these other hubs and allows us to manage them from a higher level yet?

That’s a great concept. It’s very clean. It’s very simple. Who’s going to do it? SAP is not going to do it, because they only want people to use SAP stuff. IBM’s not going to do it; they only want IBM. Oracle only wants Oracle. They're only going to do that.

Informatica might be able to do it, but they haven’t shown a willingness to go down that path. Kalido, if they had more revenues and more momentum behind them, they could try to do something like that. There’s room out there for some vendors like Orchestra Networks or Software AG that have really good registry indexing systems, to do something like that.

So that’s part of what we’re all goading the vendors about, sort of cajoling them with sticks, pushing them, poking them, saying, “Will somebody please solve the problem?”

In London, (at the MDM Institute’s Summit Series) we’re going to have one of the biggest SAP shops in the world, Philips. Royal Dutch Philips is one of the biggest SAP shops on the planet and they had built their own layer on top of all their SAP MDM instances, because they're such a big company. They have lighting, they had health care products, and they have computing. They have all sorts of different major product lines, multi-billion dollar family of companies.

They had all these different SAP instances running each of those businesses, yet they needed to have consolidated view of product, inventory, vendor, customer, that sort of thing, and distributors. They built it themselves, but then they built a layer on top of all their SAP because then it got too expensive for them to maintain because they're not in that business.

The earth-shattering news is they said we’re freezing all development in SAP MDM. We’re going to bring in Teradata MDM and have it as the layer across all of our MDM instances. So there’s somebody who’s decided that Teradata is the way to go. And Teradata has some registry-style capabilities, but they're not like the other registry guys. So it’ll be interesting to see how they do it, but there’s an example of a large company saying, “Hey, we’ve got to solve the problem. The vendors won’t do it, so we’ll find a way to do it, yet using a vendor product.”

So back to your original question, what is passive-aggressive data governance: It reinforces the notion that we need some active, integrated, proactive data governance to manage the design and the execution of our MDM systems so that they’ll be sustainable and economical.