Barcelona, Spain
October 25-28, 2016

Event Details

Please note: All times listed below are in Central Time Zone


Arrebol: Easy and Efficient Execution of Bag-of-Task Applications on Federated Clouds

We present Arrebol, a service for the easy execution of Bag-of-tasks (BoT) applications in private, public and hybrid clouds. Arrebol provides an intuitive scripting language to define large BoT jobs in a succinct way. Using a command line interface, users can submit jobs to the Arrebol submission service. It parses the jobs and interacts with the underlying cloud to create the infrastructure required to run them. Then, it submits tasks to execute in parallel, and dismantles the infrastructure when it is no longer needed, making the management of the execution of jobs completely transparent to the users. BoT applications are ideally suited to be executed over elastic cloud infrastructures. We also discuss how Arrebol leverages Fogbow, a middleware for federating clouds, to increase the capacity of the infrastructure by exploiting different hybrid settings, including cloudbursting and cloud federation. Both Arrebol and Fogbow are available as open source under Apache 2.0 license.


What can I expect to learn?

Attendees with interest in executing large BoT jobs on a cloud infrastructure will learn how Arrebol can be used to substantially simplify this task. They will learn how BoT jobs are described, and how Arrebol can be deployed at their local site. They will also learn how jobs can be submitted, have their progress monitored, and killed, if required. Then, they will learn Fogbow’s philosophy for cloud federation, how a Fogbow federation can be setup, and how Fogbow can be used to implement simple hybrid clouds where excess demand is automatically, but under the control of the cloud administrator, cloudburst to public clouds. They will have a clear idea of the effort required to deploy Fogbow atop an existing cloud. Finally, they will also learn how Fogbow can be used to federate opportunistic clouds, whose working nodes are shared desktops that are made available for being used in the cloud when they are deemed to be idle, or with a load that is sufficiently low.

Thursday, October 27, 4:40pm-5:20pm (2:40pm - 3:20pm UTC)
Difficulty Level: Beginner
Professor - Universidade Federal de Campina Grande
Francisco Brasileiro is a full professor at the Systems and Computing Department of the Federal University of Campina Grande (UFCG) in Brazil. His research is focused on the area of distributed computing, with focus on grid and cloud computing and support for escience applications. An updated list of publications can be found at http://dblp.uni-trier.de/pers/hd/b/Brasileiro:Francisco_Vilar. He... FULL PROFILE
Universidade Federal de Campina Grande
Andrey Brito is a Professor at UFCG (Universidade Federal de Campina Grande) in the Computer Science Department. Andrey’s main interests are dependability and scalability aspects of distributed systems. He coordinates R&D teams that have been contributing to OpenStack, specially regarding federation, hierarchical multitenancy, confidential computing, and integration between Ironic and... FULL PROFILE