Originating Author: Dave Vellante
Classifying data is a necessary first step in creating a cost-effective and flexible storage infrastructure. Different classes of data require different device characteristics and storage service levels but accurately classifying data and adjusting to changes in data attributes is challenging. The key is effectively combining process and technology and importantly separating the logical data characteristics (at the application level) from the physical placement of data on devices.
Many organizations lack the discipline to get started and trust that going through the process will result in continuous improvements over time. A challenge remains organizational inertia and providing responsive and even anticipatory data movement capabilities. The other vexing issue is that no matter what, classifications will not be perfect. This underscores the importance for users to take an application view of data classification.
Action Item: Taking an application view of data is the most sensible approach to data classification. Storage admin should group applications into 'chunks,' understand the data that supports applications and the business processes that rely on this data. Key criteria to evaluate should include application usage, impact of an outage and perceived value to the business by users.