HP StoreVirtual 4000 Storage - Manual (Page 31)

NOTE: If you plan on using clusters with only a single storage system, use RAID 6 to ensure data
redundancy within that storage system.
Using Network RAID in a cluster
A cluster is a group of storage systems across which data can be protected by using Network
RAID. Network RAID protects against the failure of a RAID disk set within a storage system, failure
of an entire storage system or external failures like networking or power. For example, if an entire
storage system in a cluster becomes unavailable, data reads and writes continue because the
missing data can be obtained from the other storage systems.
Using disk RAID with Network RAID in a cluster
Always use Network RAID in a cluster to protect volumes across storage systems. The redundancy
provided by RAID 1+0, RAID 5, or RAID 6 ensures availability at the storage system level. Using
Network RAID for volumes in a cluster ensures availability at the cluster level. For example:
•
Using Network RAID, up to three copies of a volume can be created on a cluster of three
storage systems. The Network RAID configuration ensures that two of the three storage systems
can go offline and the volume is still accessible, if quorum can be maintained or regained.
•
Configuring RAID 1+0 on these storage systems means that each of these three copies of the
volume is stored on two disks within the storage system, for a total of six copies of each volume.
For a 50 GB volume, 300 GB of disk capacity is used.
RAID 5 uses less disk capacity than RAID 1+0, so it can be combined with Network RAID and still
use capacity efficiently. One benefit of configuring RAID 5 in storage systems that use Network
RAID in a cluster is that if a single disk goes down, the data on that storage system can be rebuilt
using RAID instead of requiring a complete copy from another storage system in the cluster.
Rebuilding the disks within a single set is faster and creates less of a performance impact to
applications accessing data than copying data from another storage system in the cluster.
RAID 6 provides similar space benefits to RAID 5, with the additional protection of being able to
survive the loss of up to two drives.
NOTE: If you are protecting volumes across a cluster, configuring the storage system for RAID
1+0 consumes half the capacity of the storage system. Configuring the storage system for RAID 5
provides redundancy within each storage system while allowing most of the disk capacity to be
used for data storage. RAID 6 provides greater redundancy on a single storage system, but
consumes more disk space than RAID 5.
Table 8 (page 31) summarizes the differences in data availability and safety of the different RAID
levels on standalone storage systems compared with those RAID levels with Network RAID configured
volumes in a cluster.
Table 8 Data availability and safety in RAID configurations
Data availability if an entire individual
storage system fails or if network
connection to a storage system is lost Data safety and availability during
disk failure Configuration
No No Standalone storage systems with no
RAID configured
No Yes. In any configuration, 1 disk per
mirrored pair can fail. Standalone storage systems, RAID
1+0, and RAID 1+0 + spare
No Yes, for 1 disk per array Standalone storage systems, RAID 5
No Yes, for 2 disks per array Standalone storage systems, RAID 6
Planning the RAID configuration 31