Reliable self-deployment of cloud applications

Science of Computer Programming

Palma

2017

Self Cite

Cloud computing allows the delivery of on-demand computing resources over the internet on a pay-for-use basis. From a technical point of view, cloud applications usually consist of several software components deployed on remote virtual machines. Managing such applications is a challenging problem because manual administration is no longer realistic for these complex distributed systems. Thus, autonomic computing is a promising solution for monitoring and updating these applications automatically. This is achieved through the automation of administration functions and the use of control loops called autonomic managers. An autonomic manager observes the environment, detects changes, and reconfigures dynamically the application. Multiple autonomic managers can be deployed in the same system and must make consistent decisions. Using them without coordination may lead to inconsistencies and error-prone situations. In this article, we first present a simple language for expressing coordination constraints given a set of autonomic managers. Second, given a coordination expression written with that language, we propose new synthesis techniques for automatically generating an asynchronous controller. These synthesis techniques work in two steps by successively generating a model of the controller and a Java object corresponding to this model. This Java code is finally used for deploying the generated controller. As far as evaluation is concerned, we validated our approach by using it for coordinating real-world cloud applications.

Section: Related Workmentioning

confidence: 99%

Asynchronous synthesis techniques for coordinating autonomic managers in the cloud

Abid

Science of Computer Programming

Palma

2017

Self Cite

“…A preliminary version of this work was published in and was extended here as follows: We illustrate our contributions on a more complex case study. We describe more precisely the part of the protocol dedicated to the failure detection and management. We present the formal specification and verification of the deployment protocol with more details. In addition, we show a larger variety of experimental results and comment on issues detected using these analysis techniques. We present with more details the implementation of the protocol and show evaluation results in terms of performance and scalability on several real‐world applications. We present an extended discussion comparing our approach with related work, particularly those handling failures. …”

Section: Related Workmentioning

confidence: 99%

Reliable self-deployment of distributed cloud applications

Etchevers

Boyer

et al. 2016

Softw. Pract. Exper.

Self Cite

SUMMARYCloud applications consist of a set of interconnected software elements distributed over several virtual machines, themselves hosted on remote physical servers. Most existing solutions for deploying such applications require human intervention to configure parts of the system, do not conform to functional dependencies among elements that must be respected when starting them, and do not handle virtual machine failures that can occur when deploying an application. This paper presents a self-deployment protocol that was designed to automatically configure a set of software elements to be deployed on different virtual machines. This protocol works in a decentralized way, that is, there is no need for a centralized server. It also starts the software elements in a certain order, respecting important architectural invariants. This protocol supports virtual machine and network failures and always succeeds in deploying an application when faced with a finite number of failures. Designing such highly parallel management protocols is difficult; therefore, formal modeling techniques and verification tools were used for validation purposes. The protocol was implemented in Java and was used to deploy industrial applications.

“…The JEE multi-tier application is initially configured and deployed with a server at each tier, i.e., an Apache Web server, a Tomcat server, a MySQL proxy, and a MySQL database. The initial deployment phase is automated using a dynamic management protocol allowing to connect and start the involved servers and database in the right order [1,10]. In a second step, we use jmeter to inject increasing load on the Apache server and thus to simulate the clients that send HTTP requests on the managed system.…”

Section: Experiments On Our Running Examplementioning

confidence: 99%

Asynchronous Coordination of Stateful Autonomic Managers in the Cloud

Abid

Formal Aspects of Component Software

Palma

et al. 2016

Self Cite

Cloud computing is now an omnipresent paradigm in modern programming. Cloud applications usually consist of several software components deployed on remote virtual machines. Managing such applications is a challenging problem because manual administration is no longer realistic for these complex distributed systems. Thus, autonomic computing is a promising solution for monitoring and updating these applications automatically. This is achieved through the automation of administration functions and the use of control loops called autonomic managers. An autonomic manager observes the environment, detects changes, and reconfigures dynamically the application. Multiple autonomic managers can be deployed in the same system and must make consistent decisions. Using them without coordination may lead to inconsistencies and error-prone situations. In this paper, we present our approach for coordinating stateful autonomic managers, which relies on a simple coordination language, new techniques for asynchronous controller synthesis and Java code generation. We used our approach for coordinating real-world cloud applications.