By Denos Christofi, Information Management Consultant
In the current information demanding world, organizations
have assumed a two-tier infrastructure for their Information Lifecycle
Management. In simple terms, this translates to having a primary storage
platform where all active data is stored and transacted upon and a secondary
archiving platform that data retires to after it has been determined that they
are not in use anymore. In the real world though, data is active, it becomes
inactive for a while then goes back to active again. Other data may be active
only few days or few hours a month or a quarter. This suggests that a three
tier infrastructure should be implemented which allows an application to store
data effectively in three discrete tiers and platforms, mitigate their risks,
improve their performance, lower their costs and develop an environment that
caters to a future proof Information Management system.
As depicted above, all applications have active data that is
transacted upon on regular basis. They also have data that is less frequently
accessed or referenced, yet occasionally retrieved and modified. Finally, some
data are deemed of archive value, are not to be modified and are accessed more
rarely. These data, depending on internal and mostly external mandates, cannot
be changed and should be maintained at their original form since archived. That
is one of the main reasons that a three tier infrastructure is needed. While
some data may be less active, they are modifiable on a rare basis. That set of
data should not be moved to an archive system, nor should they stay in the
primary data set. Moving the data in the archive too early create issues with
data integrity, access and duplication. Keeping these data in the primary data
set makes the primary data set too large to manage, backup and restore and too
expensive to afford. A lower Primary Storage Tier can accomodate these data,
provide sufficient performance for its access and throughput requirement
and a cost effective infrastructure that lowers overall costs. In
addition, backup rules for secondary data sets can be appropriate to the rate
of data change and risks associated with loosing that data.
Consider a 10TB SAP database that needs to be protected properly
with OR and DR solutions, nightly backups, local replicas, etc. Most likely,
less than 10% or ~1TB would be the active data set. In a properly ILM
infrastructure, a storage frame outage, data corruption or a DR incident
may require the restoration of 1TB of data to get the business up and running
quickly. In the first two scenarios, the secondary data set may be on a
different platform and not be subject to the loss of the frame or the data
corruption. In either case, getting the business up and running in
approximately 1/10th of the time is more valuable than the costs associated
with the management of a three tier infrastructure, let alone all the savings
from properly aligning data to its value.