Hardware Management - Deploying and Managing a Cloud Infrastructure - page 238

Online video archives, images, snapshots, and day-to-day backups are good candidates for
tier 2 storage.
Tier 3 Tier 3 data is data that we consider appropriate for long-term backup, such as old
financial and historical records, compliance requirements, and data such as older email
conversations that may need to be kept for long periods of time. This is high-volume data
that requires infrequent access with high storage capacity at very low cost. Various tape
and virtual tape solutions are the most suitable media for this tier.
Tier 3 data includes offline master backups, gold copies, and long-time retention backups.
Policies
A multitiered storage system can provide a quick and automated way for organizations to
move data between expensive high-performance systems to lower-performing ones through
the implementation of policies that define what sort of data will fit into each tier. For example,
most financial data would fall into the tier 2 category, but if it becomes more than a year old,
it could be moved to the tier 3 category and become historical data. The same goes for data-
base records, which lose their importance as they get older and turn into historical data for
long-term backup.
Using policies with tiered storage provides a service delivery organization with the best
solution to managing data while saving time and money to meet service-level agreements
(SLAs) with the most efficiency and the lowest possible cost.
RAID Levels
Redundant Array of Independent Disks (RAID) is a disk storage virtualization technology
used to combine multiple disk drives into a logical unit. The purpose is to introduce data
redundancy and improve performance. The key goals of RAID are reliability, availability,
performance, and capacity.
RAID has different levels, which allows data distribution in several different ways. Each
level represents a different and specific balance of redundancy and required performance.
RAID levels range from 0 to 6. RAID 1 and above provide the capability to recover from
unrecoverable sector errors as well as whole disk failure.
There are a total of seven RAID levels. However, many non-standard proprietary levels
have also evolved. We will discuss only the standard levels of RAID.
RAID-0 This level offers performance enhancement through I/O parallelism across multiple
drives. There is no data redundancy and fault tolerance with RAID-0. It has no error detec-
tion mechanism and therefore failure of one disk can cause loss of all data on the array.
RAID-1 This level offers mirroring without parity checks or striping. RAID-1 implements
mirroring by writing data identically to two or more drives. The mirrored data can then
be read by any of the drives containing the requested data. Read operations can be fast
because data can be read from the drive with lower seek and rotational latencies. However,