Right Tiering is where data is classified by actual business usage and value, and then placed on the appropriate storage tier to consistently deliver the highest possible service level at the lowest possible cost.

Our goal with Right Tiering is to optimize the placement of data on the right devices, on the right tier and at the right cost. With Right Tiering, you will realize significant cost savings while maintaining or improving end-user performance goals. In general, Right Tiering provides two main business benefits:

1. Improved SLA compliance for business users by significantly reducing I/O wait time for the most critical applications.
2. Reduced TCO by placing the right objects on the right tier based on their usage patterns (e.g. highly used “hot” data on fast devices; seldom used “cold” data on slower devices, etc.).

Right Tiering for performance enhancement using EFD

The Challenge

Database performance is a critical element in day-to-day business. Meeting database SLA compliance is a constant challenge and a key factor for successful business operation. However, today’s dynamic business environment has a direct impact on application and database performance. As a result, DBAs are facing performance problems on a daily basis that have direct impact on the business. Typically, these issues are addressed by SQL and instance tuning or by adding more hardware (e.g. more CPUs, I/O channels, etc). In many cases, the solution is costly and may only be temporary. Continued rapid data growth and the current dynamic business environment require innovative and more efficient ways to solve database performance problems.

Solving the problem with Right Tiering

Recent additions to the storage arsenal, such as Enterprise Flash Drive (EFD), provide DBAs with a powerful new way to dramatically accelerate database performance. The following methodology shows the necessary steps for successful EFD implementation:

1. Identify whether or not the database is experiencing an I/O problem
2. Identify which application(s) will gain the most from using EFD
3. Identify the right candidate database objects for migrating to EFD
4. Move the objects to the EFD
5. Track the implication of the EFD on the application/users as an on going process.

By showing application data and I/O performance metrics, DBclassify bridges the gap between infrastructure and application. This unique combination provides DBAs with a very straightforward object candidates list for migrating to EFD. DBclassify identifies the objects that need to be migrated to EFD based on usage and I/O wait information, along with powerful capabilities to focus on specific application/ business users.

Right Tiering for reducing Total Cost of Ownership

The Challenge

Today, all structured data, such as that being held in your ever-growing Data Warehouse, is assumed to be created equal, and therefore subjected to the same basic placement and retention policies. However, usually a small fraction of the data in your DW is accessed frequently by critical, high value applications and user communities. The rest is either accessed infrequently, or completely dormant. But if you are unable to classify data by usage and importance, it can be stored unnecessarily on expensive, high performance tiers or stored on lower tiers that cause your users to suffer performance degradation and associated loss in productivity.

Solving the problem with Right Tiering

Right Tiering takes into consideration actual, detailed data access and usage characteristics of the workloads serviced by the database, and classifies that data by its usage profile or “temperature”. Only data classified as mission-critical (“hot”), frequently accessed by high value applications and users, should be stored on Tier 1 (e.g. EMC Symmetrix or Tier 0, EMC Enterprise Flash Devices). Less frequently accessed, but still important, (“warm”) data with less stringent service level requirements is relegated to lower performance, higher capacity storage devices (e.g. EMC CLARiiON). Dormant (“cold”) data, typically historical or static, can be marked for archival storage.