Compaq RAID ARRAY 3000 EK-SMCPO-UG. C01 - Manual (Page 54)

2-20 RAID Array 3000 Pedestal Storage Subsystem Hardware User’s Guide
Compaq Confidential – Need to Know Required
Writer: Bob Young Project: RAID Array 3000 Pedestal Storage Subsystem Hardware User’s Guide Comments:
Part Number: EK-SMCPO-UG. C01 File Name: c-ch2 RAID Array Controller.doc Last Saved On: 12/4/00 1:51 PM
While RAID 5 is ideally suited for applications with many, small I/O
operations, there is no reason why it cannot function equally well for
applications with large, sequential I/O operations. This makes RAID 5 an
excellent all-purpose RAID level.
CAUTION: RAID 5 can withstand a single failure and handle I/O activity without
interruption in degraded mode until the failed drive is rebuilt. If a second drive
fails while the RAID set is in degraded mode, the entire RAID set will fail.
Just a Bunch of Drives (JBOD)
JBOD makes it possible to connect one or standalone disk drives to the
controller. A JBOD disk drive is not part of a redundancy group, even though
the controller assigns a redundancy group number to the drive. This number
becomes that logical unit number (LUN) that the host will use to address the
drive.
One use for JBOD is to connect a system disk drive to the controller. The drive
does not become part of a RAID set, but it is made available to the host on the
same SCSI bus as the other devices controlled by the controller.
Redundant Operation
When operating in a redundant configuration, the two controllers are linked
such that, in case of a failure, the surviving controller can access the other
controller’s cache memory and complete all operations that were in progress
when the failure occurred. The controllers support two different
configurations:
■ Active/Active: One host port is active on each controller. The other port
on each controller is passive and only used if the peer controller fails.
■ Active/Passive: Both host ports on one controller are active. The other
controller’s ports are both passive and only used if the primary
controller fails.
When one controller fails, the survivor will process all I/O requests until the
failed controller is repaired and powered on. The subsystem will then return to
its previous state (that is Active/Active or Active/Passive).