Blog: Krish Krishnan: Infrastructure

« An interesting move.... | Main | Oracle's BIG announcement »

Infrastructure - A key success factor

I have been drumming the fact that for any business intelligence or data warehouse project to be successful, you need a good and robust infrastructure. This means that you will need a robust team that can design, build and support that infrastructure. Just having a good team will not mean a good foundation for infrastructure. You will need a strong and visionary leader that can lead the infrastructure strategy and provide thought leadership for the solution.

Apart from the team and its leadership, you will need to spend a few weeks of time doing in-depth analysis of the requirements of what the solution demands from performance, data needs, user security, data movement, impact to current infrastructure and come up with a finite set of infrastructure requirements and the associated project plan to accomplish the tasks.

Organizations often find themselves painted into a corner in terms of infrastructure in business intelligence or data warehousing projects. The root cause of this is either complacency about the capabilities of the infrastructure team or allowing scope creep internally or externally in the infrastructure phases of the project or being just ignorant of the critical role of infrastructure design. The net result of such a misstep often leads to burn-outs of people in the project and sometimes causes more harm to the project.

As a rule of thumb here are a few suggestions

1. Always plan for infrastructure and the teams needed to make it happen.
2. Spend enough cycles documenting requirements for infrastructure. If no business requirements are specified, use a best practices approach to building an environment for the project.
3. Do not increase scope to such an extent that the project will fail due to unnecessary demands on the infrastructure. This is a very common behavior that I have seen.
4. Plan the infrastructure design in phases. do not design multiple phases based on a single blueprint. this is especially true in the case of databases.
5. Always leave time for performance tuning. You cannot design a perfect solution.
6. If you want to attempt to bring additional ideas for a long term solution, do not guinea-pig the current project. You will fail to achieve either step and end up with frustrated users and leadership.

There are many more things to discuss in this area, but the bottom line is if you build a weak foundation, the project will collapse, if you build a very strong foundation, you might be too late for the next steps. This is always a fine balance game and you need to know how to play it to achieve success.

Share: