HP D7171A - NetServer - LPr Manual page 5

ZKLWHýSDSHU
+ 3 ý 1 H W 6 H U Y H U ý $ F F H V V R U L H V
Internal versus External Storage
Internal storage means that the drives are located within the server's chassis. External
storage allows the storage to be located in a "storage box" that is separate from the CPU.
Servers come equipped to support a certain level of internal storage, usually selected as
functions of user needs and costs. For example, a high-end server system might support a
dozen internal hot-swap disk drives whereas an entry-level system might have only a
couple of fixed internal drives. External storage can be rack-mounted or as a pedestal.
Some high-end systems use only rack-mounted storage for greatest flexibility.
When internal storage is provided, it is typically offered with built-in SCSI controllers,
channels, and backplanes. The SCSI logic is implemented directly on the server system
Mirroring
Single Channel
D=data
Figure 3. SCSI Channel Configurations
server failover configurations. When storage devices are externally connected to the
server, they are connected through SCSI or FC cabling, using an external Input/Output
(I/O) slot. The cabling can be either fiber-optic or copper. External storage can also be
used to expand storage capacity for systems with limited internal storage. For example, in a
two-node Microsoft Cluster server configuration, all storage that is shared between the two
system nodes must be external to the server. When considering a rack-mounted storage
system, you should look for the greatest density of storage modules within the standard
EIA rack sizing since floor space is a valuable commodity. Modularity is also important
since rack space may also be used for hubs, routers, and other accessories in any
combination.
Availability and Storage
Disk drives can be organized to provide high availability and data protection, as part of the
data storage function. If the drives are organized as JBOD (just a bunch of disks), then the
system has the storage capacity of those disks but no operational alternatives if one of the
disks goes down and no protection for the data.
Alternatively, any one of several RAID (Redundant Array of Independent Disks) techniques
allow sets of drives to be logically grouped in ways which reduce the drive space overhead,
while still providing comprehensive data protection. RAID systems can be implemented
through either hardware or software.
Duplexing
D M
Dual Controllers
2 single channel controllers
1 controller, 1 channel
M=mirror
board. In some configurations, hardware is
available for either mirroring or duplexing
(see Figure 3). Both terms refer to methods
of data protection. Mirroring duplicates the
D
stored data simultaneously on two separate
M
disks. If one disk fails, the other remains
available, minimizing data loss and system
interruptions. Duplexing duplicates not only
the data on a disk but also the disk controller,
protecting the data and increasing system
availability in case of either a disk or
controller failure.
External storage is most often preferred for
5