Minimizing $$\ell _2$$ norm of flow time by starvation mitigation

The assessment of a job’s Quality of Service (QoS) often revolves around its flow time, also referred to as response time. This study delves into two fundamental objectives for scheduling jobs: the average flow time and the maximum flow time. While the Shortest Remaining Processing Time (SRPT) algorithm minimizes average flow time, it can result in job starvation, causing certain jobs to experience disproportionately long and unfair flow times. In contrast, the First-Come-First-Served (FCFS) algorithm minimizes the maximum flow time but may compromise the average flow time. To strike a balance between these two objectives, a common approach is to minimize the $$\ell _2$$ norm of flow time. We first show that SRPT and FCFS are $$O(n^{\frac{1}{2}})$$ -competitive for this problem, where n is the number of jobs. Prior to this work, no algorithm was known to achieve a competitive ratio better than SRPT and FCFS. In this paper, we use FCFS to mitigate the starvation caused by SRPT. Given a good estimate of n, we prove that this approach achieves a much better competitive ratio of $$O(n^{\frac{1}{3}})$$ . To the best of our knowledge, our results provide the first theoretical evidence that mitigating starvation in SRPT leads to a provable improvement in scheduling performance.