Redefining the Data Integration Problem Set

Phasic Systems touts a combination of software, services, and even naysaying as an effective solution for effective enterprisewide governance.

The data integration (DI) space is never dull. Just when you think you have everything figured out, someone -- an upstart or established vendor, perhaps -- comes along and reinvents the wheel, or -- more precisely -- redefines the problem set.

The lesson: no one in the DI market is sitting still. This is as true of stalwarts such as SyncSort Inc. as it is of up-and-comers such as WhereScape Inc. (Both companies kicked off ambitious new DI initiatives over the last 18 months.)

It's likewise true of DI start-up Phasic Systems Inc., which debuted last year, using TDWI's World Conference in San Diego as its launch pad.

Like most of its established competitors, Phasic claims that it's targeting a DI problem set that's either ignored or ill-served by existing tools, most of which focus on the trees and lose the proverbial forest in the process.

That forest, argues Phasic founder and CEO Geoff Malfsky, Ph.D., is the ever-elusive goal of effective data governance across an enterprise.

The trees that populate this forest -- to continue our analogy -- consist not only of the disciplines that make up the practice of data governance but of the organizational entities (business units, fiefdoms, etc.) that often act as barriers to effective governance.

It's one thing for an IBM Corp., an Informatica Corp., a SAS Institute Inc., or even a Composite Software Inc. to say that enterprise-scale data governance is elusive and problematic -- and then to tout a "solution" that aims to make it practicable.

It's quite another thing for an unknown technology vendor like Phasic to do so, but Malafsky says that Phasic isn't just a technology company. It's a services and technology company. For this reason, he maintains, its products -- which included DataStar Discovery (a data discovery and analysis tool), DataStar Unifier (its data governance component), and DataStar Adapter (its nuts-and-bolts connectivity piece) -- didn't incubate in a vacuum; it was developed and honed in practice, in the trenches.

"We're born out of about 10 years of a prior company doing consulting in the entire life cycle of data. That starts with policy and strategy analysis [and goes] all the way into doing very formal enterprise architecture and modeling business processes all the way down to reverse-engineering very large data warehouses," he explains.

The first part of Malafsky's and Phasic's pitch involves shortening the data warehouse time-to-value. "Companies are spending a lot of money, and a lot of time, and involving a lot of personnel, and they're really not getting much return on this investment [in their data warehouses]," Malafsky says.

The same goes for accelerating projects once the data warehouse is in place.

"It doesn't do any good to spend 12 months creating a set of requirements when your business has changed four times in that period. The cycle time for data engineering has to get cut from 12 or 18 months down to 40 days. Something like that is heresy to traditional data modelers," he notes.

Malafsky says Phasic's approach mixes technology and services.

Take DataStar Discovery. It's part technology solution, part room-full-of-stakeholders-hashing-things-out. The idea, says Malafsky, is to collect information from subject-matter experts, data modelers, analysts, and stakeholders. It's likewise to give folks a chance to air grievances, express doubts, or -- at least -- to feel they've been heard.

"We specifically want to get [together] into the same room management types, business types, and technical types. We specifically want naysayers. Naysayers can be your biggest asset," he asserts. "Invariably, if you produce something useful -- quickly -- those naysayers can become your strongest advocates. They can become the primary change advocates in your organization."

Discovery takes all of this raw material and uses it to generate its data model. Malafsky claims it uses "semantic similarity analysis" algorithms to enrich the model with business context. From there, Phasic's DataStar Unifier handles integration and semantic mapping. "We can integrate data from relational databases, XML files, mainframe extracts -- all with complete semantic mapping. We can quickly make those semantic associations and bring all of that together," he says. "We do not use a relational data structure. We do not use a dimensional data structure; we use an object-oriented data value store. The keys are all based on a semantic vocabulary. The semantic vocabulary is defined by the business people in the discovery module and uses business terminology."

This last is Phasic's trump card, according to Malafsky. Call it insulation -- if not complete inoculation -- against change. "Just with this, we've eliminated about 12 months off the standard cycle because any concept of any kind can be immediately represented, is traceable, and can be changed at any moment.

The vocabulary exists outside of the data structure itself," he argues. "Say you're the Governance Board. Say you change the requirements or change vocabulary. All you have to do is click a button and remap it [the change] to the data, and before the meeting is over, your data is reflecting your new requirements."

Phasic proposes an almost absurdly ambitious project. It'd be ambitious enough for IBM's InfoSphere team, and it has the whole of IBM Global Services (IGS) backing it up. Why does Malafsky think corporate customers are going to be receptive to Phasic's pitch -- particularly in Global 2000 environments, where big brand names have always had CYA cachet? After all, nobody ever got fired for buying IBM, Informatica, or Oracle Corp. -- at least not when it comes to data integration.

Except, Malafsky maintains, they should have. He says the customers Phasic hears from are sounding a lot like Peter Finch's character in the film classic Network: mad as hell, frustrated with the status quo, and determined not to take it anymore.

"Companies have reached a boiling point of frustration," he maintains. "It used to be that we ran into the brick wall -- the old, old brick wall -- of 'I'm going to hire everyone who's in the right hand side of the Gartner Magic Quadrant. We're hearing resoundingly that those products are useless for what [companies] need to do, so they're a lot more receptive. They're willing to listen."