Determining server level availability and resource allocations based on workload level availability requirements

A computer program product, comprising:

• a computer-readable storage device;
• and
• a computer-readable program code stored in the computer-readable storage device, the computer-readable program code containing instructions that are executed by a central processing unit (CPU) of a computer system to implement a method of determining availabilities of servers in multiple tiers of a workload, the method comprising the steps of:
• the computer system determining an objective function of an optimization problem which minimizes a cost of achieving a requirement of an availability of the workload that includes a plurality of redundancy groups (RGs), each RG including one or more of the servers, and each of the servers being included in exactly one corresponding RG included in the plurality of RGs, the optimization problem including the requirement of the availability of the workload, resource requirements for respective RGs included in the plurality of RGs, sets of server sizes, and sets of categories of availability, the resource requirements being based on respective relative performance estimates of the servers, the server sizes including respective numbers of central processing units (CPUs) in the servers, respective sizes of data storage devices coupled to the CPUs, and respective sizes of memories coupled to the CPUs, and the categories of availability specifying durations for which the servers are operational over a specified amount of time;
• the computer system determining an availability constraint based on a number (M) of servers out of a total number (N) of servers in a given RG being required to be available so that the given RG is available;
• the computer system determining a resource requirement based on an average resource allocation being greater than the relative performance estimates in response to the servers that are available being greater than or equal to M;
• the computer system generating a non-linear integer program by adding the availability constraint and the resource requirement to the optimization problem;
• the computer system performing a Monte Carlo approach on the non-linear integer program and in response thereto, generating feasible solutions of the optimization problem, a given feasible solution including a cloud site hosting the workload, a total number of servers in an RG, server sizes of the servers in the RG, categories of availability of the servers in the RG, and a number of servers out of the total number of servers in the RG which is required to be active for the RG to be available;
• and
• the computer system minimizing a cost of achieving the requirement of the availability of the workload by performing a Pareto analysis of the feasible solutions.

A computer system comprising:

• a central processing unit (CPU);
• a memory coupled to the CPU;
• and
• a computer-readable storage device coupled to the CPU, the storage device containing instructions that are executed by the CPU via the memory to implement a method of determining availabilities of servers in multiple tiers of a workload, the method comprising the steps of:
• the computer system determining an objective function of an optimization problem which minimizes a cost of achieving a requirement of an availability of the workload that includes a plurality of redundancy groups (RGs), each RG including one or more of the servers, and each of the servers being included in exactly one corresponding RG included in the plurality of RGs, the optimization problem including the requirement of the availability of the workload, resource requirements for respective RGs included in the plurality of RGs, sets of server sizes, and sets of categories of availability, the resource requirements being based on respective relative performance estimates of the servers, the server sizes including respective numbers of central processing units (CPUs) in the servers, respective sizes of data storage devices coupled to the CPUs, and respective sizes of memories coupled to the CPUs, and the categories of availability specifying durations for which the servers are operational over a specified amount of time;
• the computer system determining an availability constraint based on a number (M) of servers out of a total number (N) of servers in a given RG being required to be available so that the given RG is available;
• the computer system determining a resource requirement based on an average resource allocation being greater than the relative performance estimates in response to the servers that are available being greater than or equal to M;
• the computer system generating a non-linear integer program by adding the availability constraint and the resource requirement to the optimization problem;
• the computer system performing a Monte Carlo approach on the non-linear integer program and in response thereto, generating feasible solutions of the optimization problem, a given feasible solution including a cloud site hosting the workload, a total number of servers in an RG, server sizes of the servers in the RG, categories of availability of the servers in the RG, and a number of servers out of the total number of servers in the RG which is required to be active for the RG to be available;
• and
• the computer system minimizing a cost of achieving the requirement of the availability of the workload by performing a Pareto analysis of the feasible solutions.