Teradata gets data warehouses ready for the long haul

Consulting services maintain peak performance for customers.

by Patric Helmaan

There's more to achieving optimum data warehouse performance than simply fine-tuning hardware or software. The entire data warehouse system must be treated in a holistic manner—looking at not only the hardware and software, but the workload and the impact of future growth as well. Teradata's Performance and Capacity Services are specifically designed to treat the entire system and provide a road map for future growth.

A data warehouse system is like a human body; it requires a well-balanced diet in order to function properly. If we do not eat enough fruits and vegetables, then we will likely gain weight, lose strength and suffer other ill health effects. If we do not eat enough protein, we will lose muscle mass and run the risk of becoming anemic, both of which can reduce our energy levels.

A data warehouse system is no different. Adding hardware can provide plenty of processing power—but in order to get the maximum benefit of the newly added hardware you must also balance the workload. Similarly, finely tuned applications won't do much good if the workload isn't balanced and there isn't enough processing capability.

Just as a well-balanced diet nourishes our bodies, an efficient data warehouse nourishes all the important aspects of the system: hardware, software, workload management and planning for future demands.

Knowing this, Teradata has developed an integrated service offer that will truly maximize the performance of the data warehouse and equip customers for the long term.

The result is something special, even for Teradata. While the offer is comprised of four distinct elements, Performance and Capacity Services looks at each element as part of a bigger solution that yields the greatest benefit for the customer. Every consulting engagement addresses four key areas:

• Application performance
• System peformance
• Workload management
• Capacity planning

Keeping the speed
Teradata's Application Performance Service addresses the design, implementation and operation of applications that enable the data warehouse to serve business users and processes.

Through monitoring and measurement, actual and potential application bottlenecks are identified, providing strategies to eliminate and avoid anything that might hinder performance.

This is accomplished using tools such as Value List Compression, which can reduce data space by as much as 30% (improving query processing time by up to 30%); Partitioned Primary Index, which enables row elimination of full table scan queries where appropriate; Materialized Views, which can pre-join data from multiple tables to dramatically improve query response times. Detailed performance improvement reports—with recommendations for further or ongoing actions—analyze application tuning and its results.

More than just hardware
System Performance Service engagements keep a constant eye on delivering the top level of hardware efficiency, resulting in both the highest levels of business effectiveness and the greatest return on a data warehousing investment.

As your business grows, the amount of information flowing to the data warehouse will grow almost exponentially. To accommodate this growth, the data warehouse will need not only to expand, but will also be required to handle greater demands in the form of more users, more queries and, in all likelihood, more complex queries—all of which require processing power. Even though achieving the best balance and performance of hardware to meet expanding needs and demands is at the heart of the System Performance Service engagement, the emphasis is on more than just hardware.

As your business grows, the amount of information flowing to the data warehouse will grow almost exponentially. To accommodate this growth, the data warehouse will need not only to expand, but will also be required to handle greater demands in the form of more users, more queries and, in all likelihood, more complex queries—all of which require processing power. Even though achieving the best balance and performance of hardware to meet expanding needs and demands is at the heart of the System Performance Service engagement, the emphasis is on more than just hardware.

As your business grows, the amount of information flowing to the data warehouse will grow almost exponentially. To accommodate this growth, the data warehouse will need not only to expand, but will also be required to handle greater demands in the form of more users, more queries and, in all likelihood, more complex queries—all of which require processing power. Even though achieving the best balance and performance of hardware to meet expanding needs and demands is at the heart of the System Performance Service engagement, the emphasis is on more than just hardware.

"Sufficient disk capacity and CPU power are only part of the engagement," says Kevin Lewis, director of Teradata Customer Service Offer Management. "The whole point of the data warehouse is to meet the customer's service-level agreements for given business priorities. As a result, our attention is on data warehouse performance as it can impact business operations."

That attention spans multiple areas. Thorough system performance audits and assessments are performed, with careful attention paid to ResUsage and AMPUsage levels within the system. Performance "best practice" criteria and baseline performance queries are defined, developed and deployed. A system for logging environmental changes is put in place and appropriate performance monitoring tools are recommended.

Achieving balance
Workload Management Service is centered on achieving the proper balance of strategic analytics and day-to-day, often minute-to-minute, tactical operations. Supporting a mixed workload environment requires a detailed understanding and tracking of query traffic. That requirement results in the development of a customized set of rules designed to provide the precise response times demanded by the data warehouse's various users.

"Monitoring the system performance as different tasks are performed at different times and over different periods of time by different communities lets us see when the system is at risk of nearing maximum capacity, and see which periods are low-usage," Lewis says. "We can then determine the best business strategies for balancing and managing workloads."

The consulting engagement process pursues performance improvement by using Teradata Priority Scheduler and Teradata Dynamic Query Manager. Teradata consultants work with the customer to analyze and assess the mixed workload environment—creating an in-depth picture of data warehouse operations, demands and needs.

The process results in a clearly defined hierarchy of demand priorities, restrictions and other thresholds. With these rules in place, the data warehouse can respond to queries based on business priority at a given time or level of usage.

Achieving that balanced management is a result of the Teradata consultants' expertise with data warehouse operations coupled with their close understanding of the customer's business.

"How we prioritize data warehouse usage is always based on business rules and business logic," says Todd Higginson, director of Teradata Services Marketing. "This is why we work with customers to build out various growth scenarios, addressing business planning as well as data warehouse expansion—brainstorming scenarios, looking at those that are likeliest to take place and those that would have the largest impact on business—and allocating warehouse resources based on what we find."

Preparing for growth
Capacity Planning Service provides documented growth plans, including incremental rollout, expansion and upgrade strategies. These plans are based on a variety of factors, including data from Teradata Manager, which provides historical information about data, query and user volumes. Plans for new or expanded applications and/or corporate acquisitions are events that can have a dramatic impact on growth plans and are always included where possible.

As with each of the elements of Performance and Capacity Service, it's a holistic approach to balancing and fine-tuning the data warehouse's current operations and future growth, leveraging maximum value from maximum performance.

Optimizing performance
Although each of the components could be executed individually, Higginson believes that large benefits can be obtained from an integrated engagement.

"For one thing," he says, "the various areas share many of the same criteria and requirements. But looking at all of the elements in unison is also a much more efficient way of optimizing the system over time."

Ultimately, the end result is about optimizing the performance of the entire data warehouse system—providing the well-balanced diet required. As with any service engagement, Teradata consultants bring a core value proposition to every aspect of the relationship: focus, expertise and results.

"Each service we offer brings a unique value to the customer," Lewis says. "But they all share the same goal and they all deliver the same result: keeping the customer's Teradata Warehouse running at peak performance all the time."

Nothing less than that is acceptable—not to the customer and, just as clearly, not to Teradata's consultants. T

Patric Helmaan is a writer and consultant whose work deals with the acceleration of technology and the challenges of technological innovation.