fairshare

Fairshare, or fair share, is a principle of resource allocation and access control that aims to distribute resources—such as CPU time, memory, network bandwidth, or budget—so that each user or group receives a proportionate portion according to policy, entitlement, or contribution. The goal is to prevent dominant users or processes from exhausting shared resources and to provide predictable performance for all participants.

In computing, fair-share scheduling is widely used in multi-user and cluster environments. Each user or project is assigned a fairshare value or quota, and the scheduler allocates resources by combining current demand with these entitlements. The approach often allows unused portions to be carried forward and can adjust priorities over time to offset historical advantage. Common implementations appear in batch schedulers and resource managers used in high-performance computing, such as Sun Grid Engine, Slurm, and PBS, as well as general-purpose systems that support quotas and cgroups. In Linux, proportional resource sharing is also achieved through weighting mechanisms like CPU shares in schedulers.

In business and public policy, fair share is applied to budgeting, cost allocation, and risk sharing. It seeks to ensure that departments, projects, or participants receive an equitable portion of resources or obligations, based on predefined criteria such as need, contribution, or responsibility.

Limitations include the dependence on how fairness is defined and measured, the potential for misconfiguration, and the possibility of unintended inequities or gaming of the system.

See also proportional fairness, quotas, budget allocation, resource scheduling.