Line: Evaluating Software Applications in Unreliable Environments

Abstract

Abstract-Cloud computing has paved the way to the flexible deployment of software applications. This flexibility offers service providers a number of options to tailor their deployments to the observed and foreseen customer workloads, without incurring in large capital costs. However, cloud deployments pose novel challenges regarding application reliability and performance. Examples include managing the reliability of deployments that make use of spot instances, or coping with the performance variability caused by multiple tenants in a virtualized environment.In this paper we introduce LINE, a tool for performance and reliability analysis of software applications. LINE solves Layered Queueing Network (LQN) models, a popular class of stochastic models in software performance engineering, by setting up and solving an associated system of ordinary differential equations. A key differentiator of LINE compared to existing solvers for LQNs is that LINE incorporates a model of the environment the application operates in. This enables the modeling of reliability and performance issues such as resource failures, server breakdowns and repairs, slow start-up times, resource interference due to multi-tenancy, among others. This paper describes the LINE tool, its support for performance and reliability modeling, and illustrates its potential by comparing LINE predictions against data obtained from a cloud deployment. We also illustrate the applicability of LINE with a case study on reliability-aware resource provisioning.

OptiSpot: minimizing application deployment cost using spot cloud resources

“…These systems have been shown in [22] to be able to scale well with respect to the system size and to provide information about the distribution of the response times of the overall system in addition to the average. Moreover, the fluid-approximated models can be easily used with tools like LINE [18,23] to perform random environment analysis. Random environments are stochastic models used to describe events occurring in the environment a system operates in [6].…”

Quantifying the Impact of Replication on the Quality-of-Service in Cloud Databases

“…To achieve this, we exploit our recent extension [32] of the fluid queueing network model introduced in [14], which incorporates an environment model. The environment model is composed of a number of stages, each representing a possible configuration or condition of a system.…”

“…The environment model is composed of a number of stages, each representing a possible configuration or condition of a system. This concept was tested in [32] to model web-based applications. In this paper, we exploit the environment model for two main purposes: (i) to model how the number of active replicas changes dynamically, increasing and decreasing as part of the operation of the database cluster; and (ii) to capture when the DB master is engaged in a replica creation process, reducing its ability to process incoming queries.…”

“…We focus on these conditions as they can have a large impact on the cluster performance. In addition, we use Coxian distributions as in [32] to capture the processing times observed at the DB cluster, and we show that this feature is key to obtain accurate estimations of the response-time percentiles, which are commonly used to define SLAs. Section IV describes the model in detail.…”

“…To illustrate the methodology we focus on read requests that belong to a single class, though this can be easily generalized to many classes representing different types of reads and writes. To analyze this model, we follow [32] in using a fluid approximation based on a set of ordinary differential equations, which provides a deterministic approximation to the expected sample path of the system state. Although approximate, the fluid model has the advantage of avoiding the state-space explosion encountered in Markovian models.…”