Leaflet

Data

Mobility

Group

www.datamobilitygroup.com

Production datasets can be pared and pruned based on application-driven rules and policies using

specialized applications that perform this function. The process, also referred to as active archiving,

solves many of the performance and manageability challenges posed by contemporary databases that

now increase in volume at the rate of 40% or more per year.

For example, a storage administrator can set a policy such that data sets reaching 20GB in volume are

pruned according to access frequency. This less frequently accessed data is then migrated to secondary

disk storage or even to tape to be held in an “active” archive. If properly implemented, historical

transactions can be maintained as complete transactions, enabling the data to be accessed long after the

source application is retired.

The benets of this technique are that the data is still immediately accessible and overall performance of

the application improves yet TCO is reduced.

Conclusion

It is a given that the volume of stored data will continue to grow and the rate of growth will continue to

accelerate. Growth will be generated by the need to retain even seemingly useless data for longer periods

of time, particularly for the purpose of responding in a timely and complete manner to legal discovery

– an increasingly common and unwelcome occurrence. But growth will also be generated by new

applications that leverage new data types – RFID for example. And while data growth can’t be avoided,

it can be managed.

Storage tiering can help manage growth while holding down overall costs related to capacity and

management. However, care should be taken when architecting the tiered storage infrastructure not to

create too many tiers as this may in fact result in increased complexity and cost. On the other hand,

attributes of the storage environment that are particularly useful in a tiered storage architecture include:

• Multiple classes of storage (classes can be dened by a combination of features)

• Virtualization

• Heterogeneous, non-disruptive data migration

• Data and operational management software

• A data classication scheme