Characterizing tenant behavior for placement and crisis mitigation in multitenant DBMSs

Abstract

A multitenant database management system (DBMS) in the cloud must continuously monitor the trade-off between efficient resource sharing among multiple application databases (tenants) and their performance. Considering the scale of hundreds to thousands of tenants in such multitenant DBMSs, manual approaches for continuous monitoring are not tenable. A self-managing controller of a multitenant DBMS faces several challenges. For instance, how to characterize a tenant given its variety of workloads, how to reduce the impact of tenant colocation, and how to detect and mitigate a performance crisis where one or more tenants' desired service level objective (SLO) is not achieved. We present Delphi, a self-managing system controller for a multitenant DBMS, and Pythia, a technique to learn behavior through observation and supervision using DBMS-agnostic database level performance measures. Pythia accurately learns tenant behavior even when multiple tenants share a database process, learns good and bad tenant consolidation plans (or packings), and maintains a per-tenant history to detect behavior changes. Delphi detects performance crises, and leverages Pythia to suggests remedial actions using a hill-climbing search algorithm to identify a new tenant placement strategy to mitigate violating SLOs. Our evaluation using a variety of tenant types and workloads shows that Pythia can learn a tenant's behavior with more than 92% accuracy and learn the quality of packings with more than 86% accuracy. During a performance crisis, Delphi is able to reduce 99th percentile latencies by 80%, and can consolidate 45% more tenants than a greedy baseline, which balances tenant load without modeling tenant behavior.

PerfEnforce Overview

“…Multi-Tenant Performance Guarantees An active area of research in multi-tenant cloud DBMS systems is tenant packing [7,16,15], or how best to colocate tenants on a shared set of machines or even DBMS instances. In contrast, we focus on the independent database user who spins up a private cluster in the cloud.…”

An Efficient Approach for Multi-tenant Elastic Business Processes Management in Cloud Computing Environment

“…Other attempts such as multi-tenant relational database management system [5] or generic SaaS elasticity have been made for this problem, but there is no known work who take into account BPMS cloud migrations and cost in a multiobjective way.…”

CPU sharing techniques for performance isolation in multi-tenant relational database-as-a-service

“…Determining which tenants to co-locate can leverage approaches in workload and resource modeling [4,25]. In the event of a performance crisis, approach such as [6] can be used for load balancing to avoid repeatedly paying penalties. Furthermore, approaches such as [26] can be used to optimize the overall provider's revenue.…”

“…Applications often reason about performance at the level of query latency and/or throughput. One approach is to provide performance assurances in a multi-tenant environment where the provider models application's workload and/or resource requirements to judiciously co-locate tenants such that their performance goals are sat-isfied [4,6]. However, if the service provider does not control or restrict the tenant workloads and aims to support general SQL workloads, such workload-level assurances are hard even when the entire server is dedicated to a tenant.…”