Towards Autonomic Workload Management in DBMSs

Abstract

Workload management is the discipline of effectively managing, controlling and monitoring work flow across computing systems. It is an increasingly important requirement of database management systems (DBMSs) in view of the trends towards server consolidation and more diverse workloads. Workload management is necessary so the DBMS can be business-objective oriented, can provide efficient differentiated service at fine granularity and can maintain high utilization of resources with low management costs. We see that workload management is shifting from offline planning to online adaptation. In this paper we discuss the objectives of workload management in autonomic DBMSs and provide a framework for examining how current workload management mechanisms match up with these objectives. We then use the framework to study several mechanisms from both DBMS products and research efforts. We also propose directions for future work in the area of workload management for autonomic DBMSs.

Dynamic Query Prioritization for In-Memory Databases

“…Internal workload management systems typically control available resources, such as CPU or main memory, and assign them to queries. Niu et al [NMP09] give a more detailed overview of workload management systems for DBMS.…”

“…Schroeder et al [SHb06,SHBI + 06] propose an external queue management system that schedules queries based on defined service-levels per query-class and a number of allowed queries in the database, the so-called multiprogramming level. Niu et al [NMP09] propose a solution that manages a mixed workload of OLTP and OLAP queries by controlling the resources assigned to OLAP queries depending on the response times of OLTP queries. Krompass et al [ [GMWD09] propose mixed workload schedulers with admission control based on query run-time prediction.…”

Interaction-aware scheduling of report-generation workloads

“…We also compare our scheduling approach to an approach that relies on query optimizer cost estimates for scheduling queries. The optimizer-based approach is an adaptation of techniques proposed by Niu et al [33,34]. We describe this optimizerbased approach in Section 6, and we demonstrate in Section 9 that QShuffler outperform this approach.…”

“…The work in [33,34,48,49] also includes a general framework for workload adaptation that can handle scheduling for time-varying workloads. In addition, the different service classes can have different service level objectives, and the scheduler dynamically adjusts the timeron threshold for each service class based on a utility function that quantifies how well the different service level classes presently meet their service level objectives.…”

“…In this section, we describe a query scheduler that uses query optimizer cost estimates, and we experimentally compare against this scheduler in Section 9. The scheduler that uses query optimizer cost estimates is based on the work of Niu et al [33,34,48,49] (in particular, the query scheduler described in detail in [33,48]). The query scheduler in these works uses the query optimizer cost estimates (termed as timerons in IBM DB2) to measure the cost of the queries executing in the system.…”