Four way support for dynamic mirror service policy

A method executed in a computer system for selecting a mirror device from a plurality of mirror devices for a logical volume comprising:

• collecting activity data of said at least one logical volume;
• determining an activity level of each of said plurality of mirror devices;
• selecting one of said plurality of mirror devices to service said logical volume in accordance with a load balancing policy of said mirror devices, said logical volume being one of a plurality of logical volumes;
• performing seek minimization processing that includes swapping first and second mirror devices in accordance with predetermined criteria, said first mirror device being selected to service requests of a first logical volume and said second mirror device being selected to service requests of a second logical volume, said predetermined criteria including a first identifier associated with said first associated mirror device that does not correspond to a second identifier associated with a location on said first mirror device identifying a location at which a mirrored portion of said first logical volume is to be stored and a third identifier associated with said second mirror device that does not correspond to a fourth identifier associated with a location on said second mirror device identifying a location at which a mirrored portion of said second logical volume is to be stored, and said third identifier corresponds to said second identifier, and activity levels associated with said first and second logical volumes being approximately the same;
• using a threshold value in determining whether the activity levels of said first and second logical volumes are approximately the same;
• determining an activity level of a logical volume using statistics in connection with at least one of read operations, write operations, and prefetched data;
• and
• determining said activity level of a particular logical device as:
• (1*# of read requests+0.5*# of write requests++0.25*# of prefetched tracks)/(time of interval)
• in which, for a predetermined amount of time represented as the “time of interval”, “read requests” represents a number of read-miss accesses associated with said particular logical volume, “write requests” represents a number of write requests associated with said particular logical volume, and prefetched tracks represents a number of sequential read accesses associated with said particular logical device.

A computer program product for selecting a mirror device from a plurality of mirror devices for a logical volume comprising machine executable code for:

• collecting activity data of said at least one logical volume;
• determining an activity level of each of said plurality of mirror devices;
• selecting one of said plurality of mirror devices to service said logical volume in accordance with a load balancing policy of said mirror devices, said logical volume being one of a plurality of logical volumes;
• and
• performing seek minimization processing that includes swapping first and second mirror devices in accordance with predetermined criteria, said first mirror device being selected to service requests of a first logical volume and said second mirror device being selected to service requests of a second logical volume, said predetermined criteria including a first identifier associated with said first mirror device that does not correspond to a second identifier associated with a first portion of said first mirror device identifying a location at which a mirrored portion of said first logical volume is to be stored, a third identifier associated with said second mirror device that does not correspond to a fourth identifier associated with a second portion of said second mirror device identifying a location at which a mirrored portion of said second logical volume is to be stored, said third identifier corresponding to said second identifier.

A computer program product for selecting a mirror device from a plurality of mirror devices for a logical volume comprising machine executable code for:

• collecting activity data of said at least one logical volume;
• determining an activity level of each of said plurality of mirror devices;
• selecting one of said plurality of mirror devices to service said logical volume in accordance with a load balancing policy of said mirror devices, said logical volume being one of a plurality of logical volumes;
• and
• performing seek minimization processing that includes swapping first and second mirror devices in accordance with predetermined criteria, said first mirror device being selected to service requests of a first logical volume and said second mirror device being selected to service requests of a second logical volume, said predetermined criteria including a first identifier associated with said first associated mirror device that does not correspond to a second identifier associated with a location on said first mirror device identifying a location at which a mirrored portion of said first logical volume is to be stored, a third identifier associated with said second mirror device that does not correspond to a fourth identifier associated with a location on said second mirror device identifying a location at which a mirrored portion of said second logical volume is to be stored, and said third identifier corresponds to said second identifier, and activity levels associated with said first and second logical volumes being approximately the same;
• using a threshold value in determining whether activity levels of said first and second logical volumes are approximately the same;
• determining an activity level of a logical volume using statistics in connection with at least one of read operations, write operations, and prefetched data;
• and
• determining said activity level of a particular logical device as:
• (1*# of read requests+0.5*# of write requests++0.25*# of prefetched tracks)/(time of interval)
• in which, for a predetermined amount of time represented as the “time of interval”, “read requests” represents a number of read-miss accesses associated with said particular logical volume, “write requests” represents a number of write requests associated with said particular logical volume, and prefetched tracks represents a number of sequential read accesses associated with said particular logical device.