Click to enlarge
Harness the Full Value of Analytics
A unified data architecture leverages all data—including multi-structured types—to deliver exclusive business benefits.
What most businesses are doing with data—forging strategies, identifying trends and predicting outcomes—isn't enough; it's only skimming the surface of the deep pool of infinite business value data can deliver.
Most organizations know how to collect, store and analyze their operational data, but not the huge volumes of interactional and observational data generated by Web logs, sensors, social media sites, call centers and other sources. These multi-structured data types are often too varied and dynamic to be cost-effectively captured in a traditional data platform. Plus, certain types of processing and analysis are difficult using only SQL for analytics.
These challenges have prompted a search for ways to transform exponentially growing volumes of complex, high-velocity data into pure business insight.
Move from Data to Value
Compared to transactional data structures and schemas, interactional data, such as from Web clickstream or set-top boxes, is more dynamic. It requires a data schema that evolves more quickly and is defined "on-the-fly" at query run time. Text, audio and images present different storage and processing challenges—they have a "format" and structure, but require pre-processing before analysis
New systems such as discovery platforms and open-source projects like Apache Hadoop take advantage of new software frameworks like MapReduce and/or distributed file systems to provide the flexibility needed to process these new multi-structured data types.
Discovery platforms, such as Teradata® Aster, are used for large-scale exploration of multi-structured data. The platforms find new insights using a variety of methods including standard SQL and business intelligence (BI) tools, new SQL-MapReduce analytics and applications, and statistical tools. By supporting MapReduce processing on a relational database, Teradata Aster provides analytic flexibility and performance at scale.
Open source solutions are being deployed more often as organizations recognize the opportunity afforded by growing volumes of data. While these technologies have cost and scalability advantages, they're not always appropriate for enterprise-class data analysis. A more complete and effective solution combines the technologies to help companies gain business insight from new and existing data.
The Architectural Differentiator
A unified data architecture can capture and store a wide range of raw data sources. (See figure 1.) MapReduce can then be used to turn the raw data into usable formats, helping to fuel new insights for the business. Hadoop can be employed to capture and refine the data types with unknown initial value. With discovery platforms that implement SQL-MapReduce, the storage-independent benefits of SQL are preserved without compromising the analytic capabilities of MapReduce.
By blending the best of Hadoop, MapReduce and SQL, a unified data architecture allows users to:
- Capture and refine data from a wide variety of sources
- Perform multi-structured data preprocessing
- Develop rapid analytics
- Process embedded analytics, analyzing both relational and non-relational data
- Produce semi-structured data as output, often with metadata and heuristic analysis
- Solve new analytical workloads with reduced time to insight
- Use massively parallel storage in Hadoop to efficiently retain data
Click to enlarge
Blended Solution Merges Benefits
A unified data architecture blends Hadoop with a data discovery platform and integrated data warehouse, allowing organizations to handle the challenges of new data sources. (See "Pick the Right Analytics Solution.") And when the developer-oriented MapReduce platform is combined with the SQL-based BI tools familiar to business analysts, users can easily and intuitively perform analytics processes that would otherwise be difficult or impossible.