Approximating an Energy-Proportional DBMS by a Dynamic Cluster of Nodes

Abstract

Abstract. The most energy-efficient configuration of a single-server DBMS is the highest performing one, if we exclusively focus on specific applications where the DBMS can steadily run in the peak-performance range. However, typical DBMS activity levels-or their average system utilization-are much lower and their energy use is far from being energy proportional. Built of commodity hardware, WattDB -a distributed DBMS-runs on a cluster of computing nodes where energy proportionality is approached by dynamically adapting the cluster size. In this work, we combine our previous findings on energy-proportional storage layers and query processing into a single, transactional DBMS. We verify our vision by a series of benchmarks running OLTP and OLAP queries with varying degrees of parallelism. These experiments illustrate that WattDB dynamically adjusts to the workload present and reconfigures itself to satisfy performance demands while keeping its energy consumption at a minimum.

Hot-N-Cold model for energy aware cloud databases

“…Constraint (13) shows that the sum of the probabilities of all blocks being accessed should equal to 1, and meanwhile at least one block is accessed during ∆t.…”

“…Ursa tries to detect the hot spots in the system and re-balance these data with minimized transformation cost. Daniel Schall et al [13][14][15] designed and implemented WattDB, which is a distributed DBMS that dynamically adjusts itself switching nodes on and off to the present workload and reconfigures itself to satisfy the performance demands. Compared with our work, Ursa and WattDB try to optimize database itself in the storage level and query engine level to achieve energy saving.…”

Frequency Selection Approach for Energy Aware Cloud Database

“…Based on this idea, there are lots of research have been done in cloud systems. Schall and Härder [8], [17], [18] designed and implemented WattDB, which is a distributed DBMS that dynamically adjusts itself switching nodes on and off according to the present workload, and reconfigures itself to satisfy the performance demands. You et al [7] proposes system Ursa which scales to a large number of storage nodes and objects and aims to minimize latency and bandwidth costs during system reconfiguration.…”

“…The energy consumption for a case is denoted as E(case) and the corresponding execution time is denoted as T (case). The accuracy of a case A(case) is defined by Equation (17)…”

Dynamic physiological partitioning on a shared-nothing database cluster

“…It evolved from a cluster with statically assigned storage nodes, only enabling dynamic redistribution of database blocks [9], to a system enhancing energy efficiency by dynamically attaching and detaching (pure) processing nodes [10], whereby query processing capacity could be adjusted to the current workload. As a further evolution step, we have combined both concepts [11] to form a fully elastic DBMS, where each node could provide storage and processing capabilities to the cluster. But data is as elastic as concrete (quoted from an invited lecture of Pekka Kostamaa at ICDE 2010).…”