For a single query, Clarinet will choose one QEP from the query's QEP-set, such that its running time is minimized. For multiple contending queries, if we pick the optimal QEPs independently, they may contend for network resources. The straw-man solution is to emulate shortest-job first and iteratively determines the QEP, placement, and schedule from the set of all QEPs of all jobs. However, it may lead to resource fragmentation. The authors present a novel approach: k-Shortest Jobs First Heuristic. Clarinet will maintain a dynamic set, D, consisting of k-shortest queries. Whenever a flow belonging to a query in D is available, and the resource it needs is free, Clarinet will immediately schedule the flow on the resource[5].