This is an excerpt from the bestselling book
Oracle Grid & Real Application Clusters. To get immediate
access to the code depot of working RAC scripts, buy it
directly from the publisher and save more than 30%.
There are many other components
that help keep the storage system secure and reliable. Some of these
include:
Hot Spares
A hot spare drive is a drive
that contains no data and acts as a standby in case a drive fails.
The hot spare drive adds another level of redundancy in the RAID
Module. If a drive of similar or a smaller capacity fails, the hot
spare automatically takes over on behalf of the failed drive until
the failed drive can be replaced with another drive. Once the failed
drive is replaced, the hot spare automatically returns to a
spare-standby status. Depending on how many hot spares are
configured, a LUN (logical unit) could remain optimal and still have
several failed drives with each one being covered by a hot spare.
This method helps to keep the RAID 1, 3, or 5 LUNs functional.
Hot Plug
A hot-plug component means that
it is electrically safe to remove or add the component while the
machine is still running. Typically, the system must be rebooted
before the hot-plug component is configured into the system.
Hot Swapping
A hot-swap component can be
installed or removed by simply pulling the component out and putting
in the new one. The system will either automatically recognize the
component change and reconfigure itself as necessary or will require
user interaction to configure the system. In both cases a reboot is
not required. All hot-swappable components are also hot pluggable,
although not all hot-pluggable components are hot-swappable. The
device being removed or reinstalled is the only device that is not
operational during this process.
Multiple RAID Controllers and
Storage Processors
RAID Controllers are the units
in a storage array that control the RAID operations. They present
one or more virtual devices or LUNS to the host system and
distribute the I/O operations to the member disks. With the help of
a RAID utility, the physical disks are grouped and formed into
Logical Units (LUNS). The LUN is the basic storage unit presented to
the host system for its use.
It is also responsible for
maintaining the mapping between the virtual device and individual
disks in the array. The existence of a RAID controller makes a JBOD
array into a RAID Storage Array. RAID Controllers are also used in
parallel access arrays for improving the I/O performance. Any
failure in the RAID controller cripples the total functionality of
the storage array.
Many of the storage arrays offer
intelligent storage processors, which control the RAID activity on
the array. The storage processor is the RAID controller for the
storage array. In many arrays, there is a local cache to handle the
data movement along with storage processors. Dual caches or
battery-backed caches help with protection and provide additional
redundancy.
Power Management
One of the most important and
unanticipated threats is the loss of power to the entire storage
unit. After providing many levels of redundancy such as multiple
paths, multiple controllers, and adequate RAID levels for disk
drives, if adequate power supply redundancy is neglected, the whole
storage unit is still at risk of losing its functionality.
Therefore, it must be equipped
with redundant power supplies. Each power supply needs to have the
required connections and wattage to supply power to the entire
storage system. Where possible, use different power circuits for
different power supplies. This way the risk of losing storage unit
functionality in the entire cluster would be minimized.
High Availability with SAN
SAN-based storage systems focus
on high availability and high performance. Storage switches are the
main building blocks for the implementation of the SAN
infrastructure. SAN incorporates many redundant hardware components
to avoid disruption to I/O activity. Even though the storage system
is part of the SAN infrastructure, it can still have multiple
connections to the host system, thus providing extra redundancy.
These storage-to-host connections can be active or on standby with
the capability of automatic failover to an alternate path.
Mirroring is another technique
used in achieving high availability. SAN enables the efficient
method of mirroring the data on a peer-to-peer basis across the
fabric. Mirroring is commonly used to deploy remote sites within a
SAN environment.
Use of FC switches provides an
extremely reliable infrastructure. Many of the switches have
hot-pluggable components, such as optic modules, which can do single
port replacements, dual CPUs, and redundant power supplies. The dual
fabric nature of the SAN provides the application with an alternate
path to the data and also contributes to high availability. The dual
fabric nature also eliminates the failure point in the data path.