Apparatus for and method of multi-dimensional constraint optimization in storage system configuration

A computer-readable memory that can be used to direct a computer to formalize the storage allocation problem for a storage system having a plurality of workload units and a plurality of storage devices, the formalization comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• selecting at least one objective function;
• generating a set of constraints for allocating workload units to specific storage devices wherein workload units are assigned to storage devices based on the constraints that have been specified by the standards and characteristics;
• applying the objective function to the assignment plan in order to refine and guide the implementation of the plan;
• and
• evaluating the optimality of the assignment plan based on the objective functions selected.

An apparatus for multi-dimensional constraint optimization in a storage system having a plurality of workload units, each having a set of standards, and a plurality of storage devices, each having a set of characteristics, the apparatus comprising:

• a constraint based solver;
• an input of the standards of selected workload units to said constraint based solver;
• an input of the characteristics of selected storage devices to said constraint based solver;
• an objective function in said constraint based solver;
• means in said constraint based solver to identify constraints based on said workload unit standards and said storage device characteristics;
• and
• an output of said constraint based solver constituting a storage assignment plan based on application of the objective function to the constraints.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• generating an assignment plan for workload units and storage devices based on the standards and characteristics, the step of generating the assignment plan comprising:
• assigning each workload unit to the first storage device for which it fits;
• randomly shuffling the order of at least one of the workload units and the storage devices;
• assigning each workload unit to the first storage device for which it fits;
• comparing the first assignment results with the second assignment results;
• and
• selecting the better of the first and second assignment results;
• and evaluating the assignment plan.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• generating an assignment plan for workload units and storage devices based on the standards and characteristics, the step of generating the assignment plan comprising:
• sorting the order of at least one of the workload units and the storage devices;
• and then
• assigning each workload unit to the first storage device for which it fits;
• and evaluating the assignment plan.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• generating an assignment plan for workload units and storage devices based on the standards and characteristics, the step of generating an assignment plan further comprising the steps of:
• generating a plurality of assignment plan runs through the steps of:
• generating a first assignment plan run by assigning each workload unit to the first storage device for which it fits;
• and then
• repetitively performing the steps of:
• rearranging the order of at least one of the workload units and the storage devices;
• and
• generating an additional assignment plan run by assigning each workload unit to the first storage device for which it fits;
• comparing the plurality of assignment plan runs to one another;
• and
• selecting the best of the plurality of assignment plan runs as the assignment plan;
• and
• evaluating the assignment plan.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• selecting at least one objective function from the group consisting of minimizing the cost of the storage system, maximizing the performance of the storage system, load balancing, the size and weight of the storage device, and availability maximization;
• generating an assignment plan for allocating workload units to specific storage devices wherein workload units are assigned to storage devices based on the constraints that have been specified by the standards and characteristics;
• applying at least one objective function to the assignment plan in order to refine and guide the implementation of the plan;
• evaluating th e assignment plan;
• implementing the assignment plan;
• obtaining empirical performance measurements of the assignment plan;
• and
• re-evaluating the assignment plan.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining standards of the workload units;
• selecting the storage devices;
• determining characteristics of the storage devices;
• generating an assignment plan for workload units and storage devices based on the standards and characteristics, assigning each workload unit to the first storage device for which it fits where the fit of the work load units to the storage devices is checked using a round robin approach;
• and
• evaluating the assignment plan.

A method of multi-dimensional constraint optimization in a storage system having a plurality of workload units and a plurality of storage devices, the method comprising the steps of:

• selecting the workload units;
• determining the standards of the workload units;
• selecting the storage devices;
• determining the characteristics of the storage devices;
• selecting at least one objective function from the group consisting of minimizing the cost of the storage system, maximizing the performance of the storage system, load balancing, the size and weight of the storage device, and availability maximization;
• generating an assignment plan for allocating workload units to specific storage devices wherein workload units are assigned to storage devices based on the constraints set forth by the standards and characteristics;
• applying at least one objective function to the assignment plan in order to refine and guide the implementation of the plan;
• and
• evaluating the assignment plan based on the objective functions selected.