As IT professionals today, we work in the most challenging of times. Never before have we seen such pressures converging on our industry. We are all too familiar with Year 2000 remediation, long enterprise application implementations, Euro currency changes and rapidly changing business requirements, not to mention IT labor shortages. Furthermore, these issues have all arrived at a time when IT budgets are coming under greater management scrutiny. What then does all this mean for data warehousing? No doubt, now more than ever, businesses require the value that a data warehouse can offer. Yet, data warehouse projects are competing with other key initiatives and management priorities for scarce resources. We often find ourselves having to deliver more for the business with fewer resources to do it. The key question then is how do we, as data warehouse professionals, provide the most value for the business and how do we manage and minimize the cost of providing that value?
Managing the Customer
One of the most important aspects of delivering a successful data warehouse is establishing and managing user expectations.Most people involved in data warehousing probably have come across several data warehouse systems that have been perceived by corporate executives as failures, not because of inadequate performance or poor data quality, but because user expectations were far beyond what was delivered. Like any other service, data warehouse success or failure is not measured so much by what is delivered, but rather by whether what is provided meets, exceeds or falls short of expectations. The criteria for judging data warehouse success is not how it compares to other projects or how much was accomplished with limited resources, but whether the consumer, the business user, is satisfied with the results. Had user expectations been more carefully managed, what had been perceived as failures might have been recognized as delivering significant business value.
Establishing a Data Warehousing Business Process
This article will discuss a technique that can assist in managing user expectations and delivering successful data warehouses while minimizing administrative effort and overhead. It involves applying to data warehouse management a well-accepted IT technique used when implementing networks and transaction processing applications: the Service Level Agreement (SLA). In fact, SLAs applied to data warehouses can be nicely extended to become an ongoing process that assists in managing and growing the value the data warehouse delivers to the business community it serves.
An SLA is a negotiated document specifying performance of a system or application. It is an explicit agreement between the business and IT that is comprised of service expectations and measurable performance criteria. As a structured agreement or contract it defines a process for enabling IT to do "business with the business." Like any business contract, an SLA specifies explicit requirements and enables those requirements to be objectively satisfied. As a management tool, it allows IT and business management to jointly manage agreed-upon objectives, measure their progress and identify at an early stage where the process may be breaking and in need of some repair. And, as an effective management tool, an SLA can go a long way in establishing and successfully managing user expectations and determining ongoing system usage patterns and changing user requirements.
The first benefit that the data warehouse SLA process can provide is an effective vehicle for communicating and confirming the business objectives of any data warehouse project. The SLA establishes explicit and documented expectations on the part of business users and documented performance and operational requirements for IT to manage. The more detailed and precise these requirements are, the more effective they will be. Documented expectations and objective measures of performance avoid the pitfalls of dealing with misperception and misunderstanding down the road.
SLA performance criteria should be based on business success requirements. Here are some examples of explicit performance criteria:
- 50 concurrent queries processed with an average query time of no more than five minutes.
- Less than four hours of planned downtime per week.
- Less than six hours of unplanned downtime per month.
- Data refreshed weekly.
These performance requirements differ by industry and the particular business supported by the data warehouse. A data warehouse supporting a financial services firm's daily pricing analysis may require much different performance characteristics than one supporting marketing analysis for a consumer packaged goods company.
Joining User and IT Perspectives
It is sometimes amazing how far removed IT can be from the business users and how far removed business people are from the realities of IT. It boils down to a good communication process.The SLA process formalizes communication between business and IT.It insures that terms are not being misinterpreted and that expectations are not being unrealistically inflated. Here is an example of an SLA response time requirement stated in the context of IT and then in terms business users can relate to:
- IT wants resources consumed below agreed-upon resource limits, based on query performance and data volume requirements.
- The users want a guaranteed response time below an agreed-upon threshold, based on the required volume of queries and data accessed.
- Here are other requirements as seen from the two perspectives:
- IT wants an adequately funded data warehouse given performance demands.
- Users want effective use of the data warehouse for the lowest cost.
- What IT requires is sufficient scheduled downtime(and allowances for unscheduled downtime) to perform maintenance.
- What end users require is sufficient uptime to derive the necessary business value from the data warehouse.
These are just some generalized examples of how SLAs can be written to relate to both sides of the data warehouse coin--service users and service providers.
Technical Constraints and Economic Realities
The foundation of the SLA process is negotiation. Like any business negotiation, it deals with value and operational as well as economic constraints. It is critical to the SLA process that the business users explicitly document their needs based on success requirements and that those requirements are then evaluated in the context of technical feasibility and operational costs. This thoughtful cost/benefit process insures that user requirements are technically feasible and worth the cost of implementing. Both IT and business management should participate in this negotiation, as well as the data warehouse implementation staff, including data warehouse architects, application administrators, system managers and DBAs.
IT management should be represented by someone who can understand business needs in the context of technical and operational constraints. Business management should be represented by someone who understands what the technical and operational constraints mean to the business. These shared perspectives enable a reasoned and economic solution to be derived with both sides fairly represented and involved. Most important in this process is that business management is involved in making trade-off decisions based on an understanding on a functional level of the technical and operational limitations.
What has been described is a process that addresses business needs while dealing with operational/technical and budgetary constraints. Let's look at a high availability requirement in a data warehouse. Suppose that the users insist on a 24x7 fault tolerant system to support a worldwide sales data warehouse. But when they learn that the fault tolerant requirement will cost an incremental $3 million, they accept a more cost-effective "compromise" that gives them an adequate level of availability for a more acceptable price. An economic trade-off occurs. In the context of technical feasibility and budget limits, resources can be effectively and realistically allocated. The SLA process forces evaluation of what is not really required to ultimately deliver what is really required given a limited set of resources.
SLAs can be a pivotal management tool during the development, roll-out and production stages of the data warehouse. In fact, the best SLAs become the impetus for continuing dialogue between the technical and business communities, helping to keep the needs of the users always in balance with the efforts of IT.
Making the SLA Process a System
The goal of the SLA process is to improve data warehouse service levels and value, to use resources in an efficient way and,importantly, not to increase IT workload and administrative overhead. To achieve these goals, companies have implemented the SLA process as a key part of the systems management infrastructure. This means automating and integrating the SLA process as much as possible. Components of an SLA system include:
- Data warehouse usage monitoring as a foundation for SLA reporting and analysis.
- A flexible SLA reporting system that explicitly measures and reports on all negotiated SLA metrics to reflect performance reality.
- Usage analysis for identifying and fixing performance issues before they impact the SLA.
Some companies are also integrating optimization techniques that better manage performance within specified SLA thresholds and that actively automate communication with data warehouse users to facilitate effective information use and manage user expectations and perceptions on an ongoing basis.
An SLA system can provide continual diagnosis of the state of data warehouse use in terms of query response time, downtime, query complexity and can determine if SLA obligations are not being met. It will also warn IT when usage patterns begin to change, signifying when adjustments may need to be made to the data warehouse and possible changes or additions made to the SLA.
Integrating SLA with Usage Chargeback
P>With the SLA process in place, users are already aware of both technical as well as economic constraints of the data warehouse that is being or has been built. In many cases, users are already funding the data warehouse in one way or another. A natural extension of the SLA process is a usage-based chargeback system.Chargeback systems can help allocate usage charges fairly across user groups and can further inform users of the actual economic cost of their information use. When integrated into the SLA process, chargeback policies and fees can be negotiated and established in partnership with the users. A chargeback system will enable the warehouse to remain appropriately funded as usage escalates. Regular chargeback reporting will give the end users and IT the tools to assess the cost and benefits of the warehouse and to continue to make more informed resource allocation decisions based on actual use and costs. Chargeback systems, when perceived as fair and understandable by the user community (especially if they have participated in establishing them), can naturally help keep resource use efficient. As we all have seen, people tend to value and use efficiently what they pay for.
Adapting to Dynamic Business Requirements
Automating and integrating the SLA management system establishes the proper IT disciplines for managing SLA for data warehouse success. It establishes a systematic process for managing end-user expectation and establishes a constructive and managed communication with the business community. It also, as much as possible, reduces the administrative overhead for managing SLAs and the data warehouse and can improve service performance and user satisfaction. In addition, it establishes a framework for adjusting to changes. Remember, it's not the changes that will kill you. Rather, it is the lack of adjustment to them that is fatal.
SLA is an iterative process; business demands change as competitive landscape changes, as user information requirements change, as usage patterns of the data change and, of course, as budget constraints change, expanding and relaxing across economic cycles and business priorities. The SLA is a living document that reflects the truth that a data warehouse is always a work in progress that supports increasingly dynamic businesses.
A Value Agreement
The data warehouse is a critical asset and should be managed like any other strategic corporate resource. The SLA is a structured management technique that can be viewed as an ongoing value agreement between the business users and IT. A specification for measuring performance and success, SLAs reduce the risk of disappointed customers, assist in establishing effective customer management and building high levels of customer service. Finally SLAs help keep the data warehouse funded and synchronized to ever-changing business needs.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access