We've all heard the expression, "you can never have too much of a good thing." But we all know that's not quite true. For example, a little dessert is good. But too much can be a problem for your waistline.

It seems data marts also fit this pattern. A few data marts can be very helpful, but too many create a huge total cost of ownership (TCO) burden.

Fortunately, with data virtualization, you can turn physical data marts into virtual ones. And when you do, you will never have to worry about having too much of a good thing.

What's Great about Data Marts

Data marts were developed as a complement to enterprise data warehouses. Typically subject or domain specific, and derivative of the warehouse, they provide a number of benefits including:

• Focused Content-Narrowing the scope to a specific domain such as finance or sales simplifies reporting and analysis.
• Query Performance-Offloading workload from the enterprise data warehouse can improve query performance.
• Data Structure -Certain reporting tools require certain structures, for example star schemas. Data marts can easily be modeled based on these structures as an alternative to the warehouse schema.
• Local Control-Users find it easier to control and modify data marts than larger warehouses.

Costs Can Outweigh the Benefits

Given the benefits cited above, data marts have proliferated rapidly. Unfortunately, as with deserts, "A moment on the lips can be a lifetime on the hips." Data mart TCO is huge. Costs include:

• Development Costs-Each data mart requires a full design, development and deployment effort.
• Operating Costs-Not only does the data need to be refreshed regularly, all the underlying databases, database servers, ETLs and more must be monitored and tuned.
• Change Management Costs-Adding new data to respond to business change requires extensive rebuilding of complex data mart schemas and ETL scripts, adding costs and reducing agility.
• Data Governance and Quality Costs-Because data is physically replicated in each data mart, each mart requires data governance to ensure consistent quality.

Data Virtualization to the Rescue

As an alternative to physical data marts, many organizations now use data virtualization middleware such as the Cisco Data Virtualization Suite, to create virtual data marts. Virtual data marts provide all the benefits listed above with far lower costs.

• Development Costs-Virtual data marts have far fewer moving parts, which lessen design, development and deployment efforts.
• Operating Costs-Fewer moving parts also means less infrastructure to maintain.
• Change Management Costs-Adding new data to respond to business change can be done in minutes or hours via virtualized data sets, rather than days or weeks in the physical data mart world.
• Data Governance and Quality Costs-With data virtualization, data mart content can be centrally governed to ensure consistent quality wherever that data is used.

Try the Data Mart Diet 

If you agree that it makes sense to lighten up on data marts, the question is how?  In other words, what is the "Data Mart Diet?"

Fortunately Rick van der Lans, data virtualization's leading independent analyst, has created the perfect Data Mart Diet program in his latest data virtualization white paper, "Migrating to Virtual Data Marts using Data Virtualization."

This whitepaper include a step-by-step approach for migrating physical data marts to virtual data marts using Cisco Information Server. Steps include:

1. Recreating Physical Data Marts as Virtual Data Marts
2. Improving Query Performance on Virtual Data Marts
3. Identifying Common Specifications Among Virtual Data Marts
4. Redirecting Reports to Access Virtual Data Marts
5. Extracting Definitions from the Reporting Tools
6. Defining Security Rules
7. Adding External Data to Virtual Data Marts

This guidance, along with the cost-benefit summary included at the start of the paper, make this paper a must read for organizations who are seeking a data mart diet.

Join the Conversation

Follow us @CiscoDataVirt.

Learn More from My Colleagues

Check out the blogs of Mala Anand, Mike Flannagan and Nicola Villa to learn more.