You, no doubt, try to incorporate best practices when engaging in database projects. This week, I was asked to present at the Oracle BIWA SIG conference on "Unusual Oracle Data Warehouse Deployment Strategies" and developed a lengthy list of worst practices in vendor evaluations, purchasing, development, deployment, management, and other areas. Many of these "worst practices" are found in all kinds of database projects. So, here is my attempt at naming ten of the worst I have observed. Any similarities to these strategies and common practices you might see on a daily basis are a coincidence... maybe...
- When benchmarking for performance, follow vendor guidelines that do not match your future real production needs. (If you follow this advice, chances are you will not meet your production performance goals later. If you are going through the work of a benchmark, you should stay engaged in the process and make sure you fully understand the testing process and how it relates to how and what you are deploying. You'll gain peace of mind and possibly eliminate a future expensive upgrade.)
- Pay no attention to what your lines of business are telling you. You and IT are the technical experts. (Maybe you are the technical expert, but my observation is that most projects that flounder today are most often because of disappointment within the lines of business and withdrawal of budget. You also want these guys partly setting priorities especially where projects could be delayed, since you would like to share that responsibility.)
- Consultants and analysts will make entirely objective recommendations. (Of course, consultants and analysts will tell you more about solutions they have familiarity with and, in the case of consultants, may lead you to solutions where they have available skills "on the bench" that they would like to sell to you.)
- Don't worry about how open the solution you are choosing is. (You will worry about a proprietary solution later when facing upgrades that are not price competitive or trying to fit complementary technologies into your deployment strategy.)
- Let procurement people choose the right servers and storage. (Procurement can get aggressive price quotes from vendors, but they might also unknowingly allow them to substitute products that are not technically equivalent resulting in wildly unbalanced hardware configurations. The resulting platform could underperform to such a degree as to fail to meet basic business requirements.)
- Pay no attention to scope and speed of deployment. (In most organizations, lines of business need solutions quickly and are more likely to continue funding where partial solutions are delivered at regular increments such that return on investment can be demonstrated.)
- Data models and applications will solve all of your needs with no modification. (Most organizations have unique business methodologies and data sources that need to be incorporated into the solution. Not accounting for this can result in missed milestones and business dissatisfaction that could kill the project.)
- Assume data quality is not an issue. (Bad data makes reports from applications and business intelligence solutions suspect. Getting to a version of the truth might not be possible. Data quality should be a concern in any system where data is entered as well as any reporting system.)
- More Database Administrators is a good thing for a project. (Sorry, but some of the worst database projects I have seen had the most DBAs allocated resulting in endless changes and disagreements. Just a very few knowledgeable DBAs with clear roles seem to work best. Further, business sponsors are much happier and more likely to fund new projects.)
- Pay no attention to where your project sponsors report in your organization. (Politics can kill your project, especially if your sponsor doesn't have the right level of influence. If their influence is on the wane, you need to find other sponsors who can take their place as needed.)
You might feel the previous list is pretty obvious. Yet, I continue to see these practices far more often than you might believe. Also, I left out quite a few, possibly some that will leap to your mind immediately. Feel free to add your favorites to the list by responding based on your experiences.