@phdthesis{Sadeghi2015, type = {Master Thesis}, author = {Mehdi Sadeghi}, title = {Distributed Flow Control and Intelligent Data Transfer in High Performance Computing Networks}, address = {Offenburg}, url = {https://nbn-resolving.org/urn:nbn:de:bsz:ofb1-opus4-9621}, pages = {x, 61}, year = {2015}, abstract = {This document contains my master thesis report, including problem definition, requirements, problem analysis, review of current state of the art, proposed solution, designed prototype, discussions and conclusion. During this work we propose a collaborative solution to run different types of operations in a broker-less network without relying on a central orchestrator. Based on our requirements, we define and analyze a number of scenarios. Then we design a solution to address those scenarios using a distributed workflow management approach. We explain how we break a complicated operation into simpler parts and how we manage it in a non-blocking and distributed way. Then we show how we asynchronously launch them on the network and how we collect and aggregate results. Later on we introduce our prototype which demonstrates the proposed design.}, language = {en} }