How to keep your database healthy
Like a new house or car, a database begins to fall apart the day it is completed. To help keep your database healthy, I've selected three simple maintenance ideas that will help maintain the longevity of your database.
Name a data czar. A tremendous amount of care and attention goes into data quality during the development of a new database. And for some time after it goes live, there is still frequent checking of the data to make sure that update processes are working properly. But once everyone is reasonably confident in the data, attention is likely to wane and errors will begin to creep into the database.
To avoid this, make checking data quality someone's job. That person should check key values after each database update and flag any anomalies. They should also have a formal mechanism to report problems, and the organization should be committed to fixing them quickly. Be sure to reward the person doing this job: either informally through public praise, or as part of their performance-related compensation. Fortunately, if this job is performed well, it should consume relatively little time.
Ditch the boat anchors. Even the biggest database servers eventually become outdated and slow. The old answer used to be replacing a server with an even bigger (and more expensive) one. It may be time to throw the big iron boat anchors overboard.
Fortunately, today's hardware and database technology make upgrading easier. In particular, something called clustering technology is now a mainstream option. In a clustered environment, smaller servers are harnessed together to manage a database. The database management software controls the separate physical servers as if they were a much larger machine.
This allows your company to buy relatively inexpensive hardware and to add servers as your need for computing power increases. If your IT group is telling you that your server is outdated, ask them about clustering as an alternative.
Stop archiving. That's right: stop throwing data away. Archiving older data is a practice from a time when disk storage was expensive. Let me give you an example: in 1999, our company purchased a two-terabyte storage array for over $1 million. In 2004, we replaced that system with another with twice the capacity for a tenth of the cost. Today, you can buy a two-terabyte storage unit for about $1,000.
Believers in archiving also argue that keeping older data will make queries run longer. But today, all of the major database systems provide the ability to partition your data in ways that avoid such problems.
Yet the practice of removing old data persists. It is also common among some service bureaus who charge a hefty premium to clients that want more than a year of history. If your internal group or a vendor insists that keeping history is too expensive, give them a lesson on Moore's law.
By paying regular attention to the condition of your data, making quick repairs when something goes wrong, and making prudent upgrades when needed, you can maintain a healthy marketing database for years to come.
David King is CEO of Fulcrum. Reach him at firstname.lastname@example.org.