Compare Mirroring to RAID-5

Resolving The Problem

What are mirroring and RAID-5, and how do they stack up against each other?

Compare Mirroring to RAID-5

Mirroring protection better protects the system from the overall viewpoint of system availability and performance. There are redundant pieces of hardware that the system can use when another one fails (availability) and, in a heavy write system environment, RAID-5 has to perform more writes to update a parity stripe than mirroring, which has to perform another write to another disk unit (performance). Also, during the time that you are running with a failed piece of hardware, mirroring is faster because the system just goes to the other disk unit, where with RAID-5 the data has to be rebuilt from the parity stripe (which also is trying to be updated to keep it up to date). Mirroring also can provide a higher level of protection because the redundancy referred to earlier can extend to duplicate buses, IOPs, disk controller, or disk drive failure. RAID-5 can only protect you from a single disk controller or disk drive failure. If a bus or IOP fails, the system goes down. The main selling point for RAID-5 is the ability to keep the system running when hardware fails for less cost than mirroring, and this is a lot better than no protection at all or the older system checksum (which is no longer supported on the RISC systems). Most of the following information about RAID-5 and the differences between it and mirroring is from the AS/400 Advanced Backup and Recovery guide, SC41-3305-01:

Device Parity Protection

Device parity protection is a hardware function that protects data from being lost because of a disk unit failure or because of damage to a disk. Data is protected by calculating and saving a parity value for each bit of data. Conceptually, the parity value is computed from the data at the same location on each of the other disk units in the parity set. When a disk failure occurs, the data on the failing unit can be reconstructed by using the saved parity value and the values of the bits in the same locations on the other disks.

Device parity protection is a high-availability function. It allows the IBM Power Systems to continue to operate when a single disk failure has occurred. The system continues to run in an exposed mode until the repair operation is complete and the data is synchronized. If a failure occurs, you should correct the problem quickly. In the unlikely event that another disk would fail, you could lose data.

The overall goal of device parity protection is to provide high availability and to protect data as inexpensively as possible. Device parity protection is built into the high-availability models of the 9337 Disk Array Subsystem, 6502, and several other Input/Output Processors (see the System Handbook for device parity capable IOPs). Device parity protection can be activated for disk units attached to one of these IOPs.

Device parity protection is not supported pre-V410 for the load source unit. It is also not available on older disk units that do not have a high-availability option.

Note: In reference to the drives and parity sets, serial numbers, type/model, resource names, number of drives per set, and the actual drives will remain constant on the system as defined. However, the parity set number can change during the IPL. The value of the parity set number is just a means of categorization.

Device parity protection is not a substitute for a backup and recovery strategy. Device parity protection can prevent your system from stopping when certain types of failures occur. It can speed your recovery process for certain types of failures. But device parity protection does not protect you from many types of failures, such as a site disaster or an operator or programmer error. The Backup and Recovery - Basic book describes how to plan a backup and recovery strategy.

For comparison information between RAID-5 and RAID-6, refer to Rochester Support Center knowledgebase document New, Benefits of RAID 5 and RAID 6 Device Parity Protection. To link to document New immediately, click herehttp://www-01.ibm.com/support/docview.wss?uid=nas8N1014458

http://www-01.ibm.com/support/icons/doclink.gif

Device Parity Protection Compared to Mirrored Protection

Mirroring is writing data to two identical disks. If one fails, the system uses the other. Mirroring is a high-availability function. It allows the system to continue when a single disk-unit failure has occurred. Data is protected from loss while the failure is being corrected unless the other disk unit in the mirrored pair also fails.

Mirroring provides the best performance after a failure because data can be used directly from the mirrored disk. Device parity protection is be used directly from the mirrored disk. Device parity protection is slower than mirroring when a failure has occurred because data on the failing unit has to be reconstructed from the data on other units.

Mirroring can also provide protection from other hardware failures. For example, if the disk units in a mirrored pair are attached to different IOP cards, an IOP failure would not impact availability. Mirroring can be done on any disk unit and model type, including the load source disk unit.

The disadvantage of mirroring is that it requires the most disk space and is the most expensive availability option. Device parity protection requires significantly less disk space than mirroring.