Teradata, which practically invented the enterprise data warehouse, has always been a company where engineering came up front. Company veterans and longtime customers knew Teradata products by their model numbers; a 6800 told you that you were using a giant enterprise data warehouse, as opposed to a 2800 that was sized for departments.
As longtime data warehouse provider, Teradata was the go-to box for complex SQL analytic queries across large bodies of data. So it's not surprising that in recent years, emergence of commodity hardware and open source infrastructure posed an existential threat. While Teradata's optimizations of its mature SQL engine, designed to handle complex queries, extended down to processor and backplane, commodity alternatives promise brute force alternatives that could be considered "good enough."
Today, Teradata provides superior service levels to Hadoop. For now, it continues to maintain a highly diverse line of workload-optimized appliances that, because of their specialized natures, will never be cost-competitive to commodity hardware. And over the long run, Hadoop will improve its performance, although using an underlying file system it will never be as efficient as true databases like Teradata, Oracle , SQL Server, DB2, or any of the wide breed of SQL columnar analytic databases.
But Teradata has a trick up its sleeve with its recently-unveiled Intelliflex architecture that makes its hardware become software-defined. While Teradata appliances are likely never to be as cheap as commodity hardware, Intelliflex provides the company a path to getting better economies of scale that will make it more price-competitive by consolidating its varied portfolio of workload-optimized specialized appliances with more standard, configurable designs.
The past year saw the company changing senior management, divesting some assets, and refocusing on several new themes that are opening access to Teradata in the cloud, and doubling down on solutions to reach beyond the company's traditional IT constituency to the business, which signs the checks.
"Teradata Everywhere" and "Borderless Analytics" are branding that signify how Teradata is making its technology more widely accessible. In part, it refers to the new unbundling of Aster as analytics software, which we reviewed a couple weeks back. It also describes Teradata's new, broadened cloud focus. It's the next step of a strategy that debuted with the Teradata Managed Cloud just over a year ago where Teradata hosts your analytic appliance in a single-tenanted deployment redolent of the old application service provider model.
Now Teradata is also making itself available in the public cloud, initially on Amazon, and later, on Microsoft Azure. Teradata is prepackaging templates for some of the most common cloud use cases, including cloud bursting (bursting on premise workloads to the cloud as safety valve); data lab (for testing new analytics use cases or scenarios); and disaster recovery (where the cloud is used for backup).
Teradata's cloud strategy is not all that unusual compared to other IT household names; IBM, Oracle, and SAP are also going there because that's where customers are headed. There are subtle and not so subtle differences in the way that each is supporting the cloud. For instance, Teradata diverges from Oracle with with a single-tenanted rather than multi-tenanted model; but both share the common thread of making deployment as transparent as possible from on premise, so you can run the same bits on both ends.
But the question is which target audience is Teradata seeking to reach and, given its IP over workload optimization, will its reputation stay intact when exposed to the open air of the Amazon cloud and commodity S3 storage? For existing Teradata customers, Amazon (and eventually Azure) presents a great place for running proof of concept and test/development workloads, or the place for stretching the analytics budget by running workloads on a project basis that would not otherwise justify the cost of buying new Teradata boxes. That represents a win-win for Teradata and its existing installed base.
But what about new customers, for whom the default choice would otherwise be Amazon Redshift? That represents a steeper challenge for Teradata, since for many of them, Redshift performance should be adequate. To be competitive, there need to be enough customers of modest means that know about designing the complex queries where Teradata shines. For now that's a very open question, but Teradata can draw hope from Oracle, whose recent quarterly results proved that there is a latent market of midsized enterprises that are anxious to run on established platforms considered the enterprise gold standard.
As for Teradata's solutions focus, it's not a huge stretch because the company already enters the starting gate with significant IP, from the broad portfolio of vertical industry logical data models that came out of its professional services engagements to the rich library of analytic functions that came with the Aster platform. For instance, Aster provides path analysis and graph modeling that is accessible through SQL functions. The building blocks are there for Teradata to build some higher analytics value-add. It's a strategy that IBM has also applied when combining assets like SPSS statistical modeling (and now Spark) with vertical industry frameworks generate through its services business.
Customer Journey, followed by Analytics of Things (for IoT) is the prototype for Teradata's emerging solutions business. Over the past year, Teradata divested Aprimo, a marketing campaign management application, but kept the analytics pieces including Real-Time Marketing (RTM) and Customer Intelligence Management (CIM). For Customer Journey, they built a real-time path analysis application that integrated Aster nPath analytics with RTM, and then added a GUI (which replaces the SQL command line of Aster nPath) so you can drill down on specific nodes in the customer's path (e.g., as they navigated through a website, contacted a call center, and/or communicate via text or email, etc.) and see in real time what's happening in the journey.
Teradata plans to build more such frameworks and applications to bolster its solutions business. Undergirding this, they have expanded the professional services teams, developed a methodology that includes "business value frameworks" that specify business functions and which solutions could be developed. For instance, with the financial industry, Teradata has identified marketing and customer experience, risk management, performance management, compliance, operations management, and infrastructure management as candidate solution areas. There are targeted, six-week jumpstart RACE engagements for developing analytic models, and based on demand, those models may be packaged as analytic frameworks and/or full-blown business solutions.
Of course, the devil for Teradata's solution transition is in the details. For starters, Teradata's services operation is still rather bifurcated: there is the mother ship and then there's several acquisitions -- ThinkBig for Big Data and Claraview for customer experience -- that are still autonomous. While ThinkBig and Claraview are well positioned for providing unique IP, the challenge is minimizing potential turf distraction and inconsistent execution.
Then there's the risk of channel conflict, as a solutions business will put Teradata in competition with many of its ISV partners, especially SAS where it has an arm's length relationship. Threading the needle will dictate that Teradata not compete with general-purpose BI or analytics vendors, but aim for vertical niches that would be narrow for them. That said, that's also likely the sweet spot for SAS.
Either way, in an era of commodity infrastructure that is gradually closing the gap with optimized engineered platforms, Teradata has little choice but to aim higher up the business value chain.