You are currently browsing the tag archive for the ‘Teradata Aster’ tag.

At its annual industry analyst summit last month and in a more recent announcement of enterprise support for parallelizing the R language on its Aster Discovery Platform, Teradata showed that it is adapting to changes in database and analytics technologies. The presentations at the conference revealed a unified approach to data architectures and value propositions in a variety of uses including the Internet of Things, digital marketing and ETL offloading. In particular, the company provided updates on the state of its business as well as how the latest version of its database platform, Teradata 15.0, is addressing customers’ needs for big data. My colleague Mark Smith covered these announcements in depth. The introduction of scalable R support was discussed at the conference but not announced publicly until late last month.

vr_Big_Data_Analytics_13_advanced_analytics_on_big_dataTeradata now has a beta release of parallelized support for R, an open source programming language used significantly in universities and growing rapidly in enterprise use. One challenge is that R relies on a single-thread, in-memory approach to analytics. Parallelization of R allows the algorithm to run on much larger data sets since it is not limited to data stored in memory. For a broader discussion of the pros and cons of R and its evolution, see my analysis. Our benchmark research shows that organizations are counting on companies such as Teradata to provide a layer of abstraction that can simplify analytics on big data architectures. More than half (54%) of advanced analytics implementations are custom built, but in the future this percentage will go down to about one in three (36%).

Teradata’s R project has three parts. The first includes a Teradata Aster R library, which supplies more than 100 prebuilt R functions that hide complexity of the in-database implementation. The algorithms cover the most common big data analytic approaches in use today, which according to our big data analytics benchmark research are classification (used by 39% of organizations), clustering (37%), regression (35%), time series (32%) and affinity analysis (29%). Some use innovative approaches available in Aster such as Teradata’s patented nPath algorithm, which is useful in areas such as digital marketing. All of these functions will receive enterprise support from Teradata, likely through its professional services team.

The second part of the project involves the R parallel constructor. This component gives analysts and data scientists tools to build their own parallel algorithms based on the entire library of open source R algorithms. The framework follows the “split, apply and combine” paradigm, which is popular among the R community. While Teradata won’t support the algorithms themselves, this tool set is a key innovation that I have not yet seen from others in the market.

Finally, the R engine has been integrated with Teradata’s SNAP integration framework. The framework provides unified access to multiple workload specific engines such as relational (SQL), graph (SQL-GR), MapReduce (SQL-MR) and statistics. This is critical since the ultimate value of analytics rests in the information itself. By tying together multiple systems, Teradata enables a variety of analytic approaches. More importantly, the data sources that can be merged into the analysis can deliver competitive advantages. For example, JSON integration, recently announced, delivers information from a plethora of connected devices and detailed Web data.

vr_Big_Data_Analytics_09_use_cases_for_big_data_analyticsTeradata is participating in industry discussions about both data management and analytics. As Mark Smith discussed, its unified approach to data architecture addresses challenges brought on competing big data platforms such as Hadoop and other NoSQL approaches like that one announced with MongoDB supporting JSON integration. These platforms access new information sources and help companies use analytics to indirectly increase revenues, reduce costs and improve operational efficiency. Analytics applied to big data serve a variety of uses, most often cross-selling and up-selling (for 38% of organizations), better understanding of individual customers (32%) and optimizing price (30%) and IT operations (24%). Teradata is active in these areas and is working in multiple industries such as financial services, retail, healthcare, communications, government, energy and utilities.

Current Teradata customers should evaluate the company’s broader analytic and platform portfolio, not just the database appliances. In the fragmented and diverse big data market, Teradata is sorting through the chaos to provide a roadmap for largest of organizations to midsized ones. The Aster Discovery Platform can put power into the hands of analysts and statisticians who need not be data scientists. Business users from various departments, but especially high-level marketing groups that need to integrate multiple data sources for operational use, should take a close look at the Teradata Aster approach.


Tony Cosentino

VP & Research Director

Unlike other recent conferences that seem to focus almost exclusively on cloud computing, this week’s Teradata Partners Conference emphasized big data and analytics. The vision that Teradata lays out is one in which new technologies such as Apache Hadoop live side by side with more traditional enterprise data warehouses (EDW) and companies have the flexibility to define their own approaches to BI tools. This approach, at least in the near and medium terms, makes a lot of sense, and is backed by our own research into big data, which shows relational databases are still the predominant tool for delivering big data analytics and solutions to the enterprise. Companies have spent a lot of money on their current infrastructures, and not many have the stomach for a rip-and-replace strategy. Nor do most organizations have the tools and the skillsets yet to take full advantage of all of the newer approaches coming into the market around big data analytics.

Now part of Teradata’s big data and analytics strategy is its integration of Aster, which the company acquired about a year ago, into the Teradata portfolio. Aster offers some big advantages for accessing big data through commonly used and understood query approaches such as SQL. In fact, the SQL-H approach that Aster pioneered allows ANSI-standard SQL exploration of big data, and so far Aster is the only product on the market with such capabilities. SQL-H allows users to employ familiar SQL approaches to the MapReduce framework and take advantage of the massively parallel processing nature of Hadoop. It does this by leveraging HCatalog, which abstracts a metadata layer from Hadoop and provides hooks for the Aster query engine. The fact that this approach uses standard SQL makes it fit in well with existing BI tools and processes.

The Teradata approach relies heavily on Hadoop, the fastest-growing open source ecosystem around big data, but one that is still in its early stages; most organizations have not even put it through a proof of concept phase. Its most mature use case, and the one where organizations seem to be deriving the most value, is one in which Hadoop acts as a supercharged landing strip and refinery for different types of data. Hadoop is very good at capturing and storing data, and at applying low-level math on an extremely large scale in a batch process. The functional tasks it performs, such as filtering, sorting, counting and averaging, are valuable in deriving order out of the chaos of unstructured data. Once analysts apply some basic structure to the data, they can apply an iterative approach to look at data in more advanced ways and develop more complex algorithms.

Last week Teradata announced the Teradata Aster Big Analytics Appliance. In a close developmental partnership with HortonWorks, the Big Analytics Appliance aligns with Hadoop via Hortonworks’ HDP 1.1 and integrates hardware with software. The system provides more than 50 prebuilt MapReduce functions that are accessible in SQL. Because it uses standardized SQL, the system can leverage off-the-shelf BI tools and current ETL deployments. Another key feature is Teradata Viewpoint, which provides server management and monitoring for Teradata’s EDW platform and Teradata Aster. Support for Hadoop is expected early 2013. The system connects the Big Analytics Appliance to the Teradata EDW platform via 40Gbps InfiniBand, SQL-H, TD-Aster and TD-Hadoop connectors.

Teradata promises enterprise-class customer support for the Big Analytics Appliance. The company provides three levels of customer support across the entire system, including Hadoop. Any calls escalated to the highest level, level four, are routed to various centers of expertise, including platform engineering, Aster engineering and Hortonworks, respectively.

This enterprise assurance factor is a key part of the Teradata’s analytics strategy and will be a key determinant of Teradata’s future success. In discussions with customers at the conference, I got a positive feeling about Teradata’s ongoing commitment to tight integration within its systems, its approach to professional services and its general approach to customer support and satisfaction. Enterprise assurance is an intangible driver of purchasing behavior, but it can be especially strong in changing times.  One challenge will be for Teradata to maintain this value as it grows. It will be incumbent upon its professional services divisions to attract and retain a high level of talent, and structure its service delivery models in such a way that any growing pains are seamless from a customer perspective.

Teradata’s Big Analytics Appliance and big data analytics strategy provide a compelling story that addresses the analytics skills and staffing gap revealed by our big data research and promises high levels of customer assurance.


Tony Cosentino

VP & Research Director

RSS Tony Cosentino’s Analyst Perspectives at Ventana Research

  • An error has occurred; the feed is probably down. Try again later.

Tony Cosentino – Twitter

Error: Twitter did not respond. Please wait a few minutes and refresh this page.


  • 73,715 hits
%d bloggers like this: