IRIS Volume Manager provides technical users with sophisticated disk management tools previously available only in mainframe OLTP environments. Through its implementation of virtual disks, IRIS Volume Manager gives system managers a wide range of options that improve the availability and performance of critical files and databases. Virtual disks eliminate the physical limitations of disk hardware by providing important features such as mirroring, striping, and spanning. The IRIS Volume Manager on-line performance monitoring and management utilities allow systems managers to tune the system for maximum performance without disrupting users or applications.
IRIS Volume Manager brings the technology of high productivity, user extensible, graphical interfaces to the world of system administration. For expert administrators, IRIS Volume Manager works hand-in-hand with familiar IRIX utilities and acts as a highly effective display for disk and file system management. For the less expert administrators, IRIS Volume Manager acts as a safe, simple, and effective guide in the world of disk subsystem management, allowing them to manage more complex configurations. The result is better informed and more effective decisions concerning disk-resource management that improve database and file system operation.
Mirroring can be done on a selective basis so that only critical volumes or portions of critical volumes need to be mirrored. This fine-grained mirroring is more economical than mirroring entire disks.
Using IRIS Volume Manager, the system administrator can also perform file system snapshot backups without interrupting service. Plexes can be brought off-line, backed up, and then updated without interrupting any applications accessing file systems associated with the plexes.
Once the failure has been repaired, IRIS Volume Manager revives the failed disk by initiating whatever disk operations are required to make the failed plex consistent with its mirrors. This recovery operation is performed without system interruption.
Since the recovery process may involve significant disk 1/0, IRIS Volume Manager allows the system administrator to specify the appropriate recovery rate given the magnitude of the data and applications involved. This flexibility can either force high-priority recoveries to happen immediately, or specify a slower pace, which will have less impact on the 1/0 bandwidth of the system. In some failure cases, IRIS Volume Manager can resolve the problem automatically by rewriting the problem subdisk after a read failure. This returns the system to normal operating condition without requiring any operator-involved repair. In all cases, errors can be logged for later review.
IRIS Volume Manager also provides mechanisms for performance analysis for system tuning. Device utilization or performance analysis can be accomplished via general 1/0 statistics or detailed logs. The analysis can involve any components of the IRIS Volume Manager system. The 1/0 statistics allow general identification of problems. A complete trace of all 1/0 activity can be done on a per volume basis for use in detailed performance studies or simulations.
An analysis can identify inefficient configurations, read/write bottlenecks, or unused resources. Examples include keeping track of read-to- write ratios and identifying hot spots where an unusually large amount of activity is taking place so that better mirroring or striping configurations can be identified. Such analysis helps the administrator determine how to tune the system to maximize disk-access performance or determine the most critical areas for high-availability protection.
Application programs operate on volumes using existing IRIX interfaces, so modifications to existing software applications are unnecessary.
IRIX file systems or other kernel components operate transparently on volumes through the block and character device switches. For example, once IRIS Volume Manager is installed, existing IRIX file systems or databases can then be easily striped or mirrored across multiple disks.
The IRIS Volume Manager environment can be easily customized via administrator control over various read/write policies, error handling policies, and the addition of volume usage types. Migration to future operating system environments is straightforward because IRIS Volume Manager is implemented primarily in the application level with minimum IRIX kernel-integration requirements.
Although IRIS Volume Manager uses default settings for how and which information is displayed, each user can customize the display for specific needs. The user has full control over the display of information such as icon sizes, textual data included, and color usage, as well as magnification and scrolling to support the display of all data associated with large configurations. IRIS Volume Manager recalls each user preference whenever that user accesses the software.
IRIS Volume Manager provides a vehicle for displaying the relationships between physical disks and logical volumes. Selecting Disks displays each physical disk in the system, and the user can easily view which volumes, plexes, and file systems are related to each physical disk.
Selecting Volumes produces the same information, but begins with file systems and volumes, and permits the user to expose the structure that supports them. Both views can be displayed in parallel.
Most important, the user can create specialized views of the disks and file systems. These user-created views may combine the physical and logical views to display information most relevant at the time. These customized views are continually maintained by IRIS Volume Manager.
Examples of configuration actions include creating the initial system, rearranging volumes on or across disks, including striping, and creating or deleting plexes (copies) of volumes. Modifications may involve using both the physical and logical views of the system simultaneously.
Modification actions also serve to recover from file or disk system problems, or to prepare for backups or archiving of data. Plexes can be brought off-line for backup and later revived to ensure consistency. All of this activity can be accomplished on-line, with no disruption of system operation.
Analysis involves any or all views of the system. Such views are by physical disk, subdisks, partitions, plexes, or volumes. The analytical data can be displayed graphically and monitored on line. The data is sampled and updated on the screen at a user-selectable rate. The user can keep track of read-to-write ratios to identify hot spots by having various icons colored differently depending on the percentage of time that the structure is busy. Such an analysis can identify inefficient configurations or unused resources and help the administrator determine how to tune the system to maximize disk-access performance or determine the most critical areas for high-availability protection.