Ten Techniques to Guarantee a Successful DW/BI Solution
December 29, 2010
The ingredients to a successful Data Warehouse / Business Intelligence deployment include good project management, effective communication and using DW/BI tools to their full potential. Denise Rogers presents her top 10 techniques to guarantee a successful DW/BI deployment.
The ingredients to a successful Data Warehouse / Business Intelligence deployment include good project management, effective communication and using DW/BI tools to their full potential. A Data Warehouse / Business Intelligence application is being built, NOT an OLTP application. So use the ETL solution and its features in support of the solution, the Business Intelligence software to its full potential, the DBMS and its Data Warehouse functionality. Other essential ingredients include getting the resources trained or experienced hires or both that know how to exploit the features and functionality that each of the components of the solution has to offer.
The mantra should be to design a solution based on the business requirements and the strategic direction, not to fit a DW/BI solution into an OLTP strategy for deployment that has been around for years.
The functionality of the tools should be completely understood, this is the responsibility of the team, such that all available components of the tools are alternatives to be explored as part of the design. This ensures a comprehensive understanding of the features and functions available to design a robust solution. So without further delay, I would like to introduce the top 10 ways to guarantee a successful DW/BI deployment.
Number 10 - Manage the DW/BI project…there are too many risks not to
A DW/BI project is huge undertaking for any organization. On this, we must agree. So why would any organization not want to have a formal project management approach to ensure success? The answer should be obvious. A formal process for administering the project ensures that communication, risks, scope, cost, time and quality are managed successfully and effectively, are required and essential.
Number 9 - Sponsorship…don't leave home without it
There's quite a bit of momentum for the DW/BI incentive. That's great! However, without the approval and sponsorship from senior management, the project goes nowhere really fast! Executive sponsorship ensures that the project moves forward in terms that it enables the alignment of resources, approved budgets and opens the lines of communications. This is gold for the DW/BI project.
Number 8 - Documentation is the gift that keeps on giving
Documentation IS the gift that keeps on giving. Why? There are so many moving parts and so many components to the DW/BI solution. Having living and breathing documented artifacts, either using standardized documentation practices or through standards created for the project enables the DW/BI solution to become a scalable, supportable and maintainable set of applications with minimal impact on resolving issues.
Number 7 - Deploying a DW/BI solution is hard enough, don't make it harder
Building a DW/BI solution is one of the most ambitious undertakings for any organization. There is no real need to take on unnecessary risks. One such risk is using buggy, uncertified software code to build parts of the solution. The beta version may be at a reduced cost at the onset but how much will it cost the project when it breaks during the production deployment? And it will break! Staying a release or two back is not a bad idea. These versions have most of the bugs worked out so there's minimal risk. A DW/BI solution should be the leading edge of software not the bleeding edge!
Number 6 - Metadata is more important than you think
Metadata management and data governance is an integral part of any DW/BI solution; it's more important than you think! Metadata not only defines and describes the data elements in the data mart, it provides a lineage or audit trail back to its source of origin. Metadata defines the purpose of the data, why it was originally created and its business owner. This is valuable information when the data in a report is being questioned during an audit and requires validation. It also reduces the time required to enhance a BI application specific to the creation of new reports.
Number 5 - A spreadsheet does not a UI make
When there has been a lot of effort invested in cleansing, transforming and loading data into a DW/BI environment and there is a high demand to report on the data, give the business a reporting solution that really speaks to their requirements. A spreadsheet is good for reporting in certain situations but if there is an enterprise BI solution with the features and functionality that the business community is asking for, it is the job of the IT team to create an application that actually uses all that the BI solution is capable of! Remember the BI reporting solution is not just a spreadsheet; the evaluation was exhaustive, all features and functionality were tested, why only use 10% of the functionality to deploy the DW/BI solution.
Number 4 - Upgrade the infrastructure…it makes DW/BI solution work…really
A DW/BI application needs to have an infrastructure that supports massive amounts of data to be moved, transformed, cleansed, loaded, stored and reported on. To ensure success of a DW/BI initiative, it is essential that the project plan includes tasks that assess and upgrade the infrastructure. A DW/BI requires lots of storage and lots of memory; make sure the platform is ready to host the solution.
Number 3 - ETL can do a lot of things...and it should
An ETL solution does so much more than extract and load data. A complete ETL solution includes data profiling, standardization, cleansing and transforming. So design an ETL solution that uses its features; make the ETL software work to 100% of its design not 10%…..ETL software is NOT just a data mover!
Number 2 - Put the Database to work
Oracle, UDB and SQL Server (to name the big 3) have come a long way in the DW/BI space. Get all the components working to support a DW/BI solution not just part of the solution; all of it! The DBMS has quite a few DW/BI features that offer star schemas with aggregate functions, data cubes with all the features that support drill through. The data mart should not be in 3rd normal form with views that supports complex joins.
Number 1 - Get the right resources in place to do the job
Some of the DW/BI biggest successes and failures are often linked to how the project has been resourced. For organizations that are embarking on their first DW/BI deployment, it is extremely important that teams are led by people that are seasoned veterans in DW/BI. These senior architects and developers will influence the organization to align its thinking related to DW/BI solution. This includes training the team to use the DW/BI software and using experienced hires to set the strategic direction and implement policies and procedures that ensure success of the DW/BI project.