Hardware Management - Deploying and Managing a Cloud Infrastructure - page 237

Organizations can establish their data storage tiering requirements by determining
functional differences in data such as restoration and replication needs. The general rule of
thumb is that dynamic data, which is often mission critical and required by applications and
users quickly and frequently, is stored in lower-tier storage for increased performance and
nonessential data, such as long-term backups, can be stored in slower, higher-level tiers.
The different tiers are determined by the kind of access, performance and reliability
associated with the kind of data you are storing. Time and money can be saved by imple-
menting or making use of a tiered data storage infrastructure.
Performance Levels of Each Tier
Organizations have to define each type of data and then determine how each is classified.
The criteria can be in the form of questions, such as, How critical is the data for the daily
business processes of the organization? How long is the data going to be archived (days,
months, or even years)? How frequently will the data change and how quickly will it need
to be restored? The answer to these questions and many more will determine which tier is
appropriate.
Tier 0 Tier 0 data is defined as data that is mission-critical, frequently accessed, or recently
accessed or that requires high levels of security. This data requires very high I/O activity
and low latency and a mix of reads and writes. It should be stored in high-performance and
highly available storage systems such as disks or SSDs in a RAID configuration with parity.
Tier 1 storage systems must have the best performance, reliability, and manageability, yet
they need less capacity, which translates to being the most expensive storage systems.
Tier 0 storage would be appropriate for online look-up tables, VM/VDI (virtual machine)
files, indices, databases, and log or journal files.
Tier 1 Tier 1 data includes data generated by major business applications such as customer
relationship management (CRM) and enterprise resource planning (ERP) software as well
as email and other documentation data. This type of data has a higher ratio of storage to
I/O activity and has only moderate I/O activity with a mixture of reads and writes as well as
some random and sequential access and writes. This tier hovers between performance and
cost and should be used for data that does not require absolute microsecond response times
yet still need to be reasonably fast and adequate for its purpose. Cheaper alternatives to
Tier 1 systems can be used, such as less-expensive disks that perform relatively well and
can be set up as RAID.
Appropriate data for tier 1 systems is similar to that of tier 0 systems—such as database
tables, VM/VDI files, and web user files—but may be considered less critical in nature.
Tier 2 Tier 2 data is still important, but it may not be required in daily business pro-
cesses, such as financial and transactional data and other machine-generated data. Tier 2
systems focus on high storage capacity at low cost with occasional quick random access to
data. The data is important, but it doesn't require high performance and high availability,
so it doesn't need to be placed in either tier 1 or tier 0 storage. High-capacity disks which
offer more storage than performance may be used.