Seminario - Performance of Large Scale Distributed Systems (21 aprile 2017 ore 10:30)

Seminario - Performance of Large Scale Distributed Systems (Issues, Trends, Problems and Solutions)
Venerdì 21 aprile 2017 ore 10:30
Sala del Consiglio DIMES
5° Piano Cubo 42C
Speaker: Prof. Eleni Karatza
Department of Informatics, Aristotle University of Thessaloniki, Greece. Questo indirizzo email è protetto dagli spambots. E' necessario abilitare JavaScript per vederlo.

Recent advances in networks and computers have led many aspects of our daily life to depend on distributed interconnected computing resources. Large scale distributed systems such as grids and clouds are used for serving large and complex applications. Grids and clouds performance became more important due to the increase of users and applications. Effective management of distributed resources is crucial to use effectively the power of these systems and achieve high system performance.

However, there are important issues that must be addressed for large scale distributed systems, such as: performance, resource allocation, efficient scheduling, energy conservation, reliability, cost, availability, quality. Furthermore, due to the cost of electrical power consumption and the environmental impact, energy efficiency in large scale systems is a global IT concern. Complex multiple-task applications may have precedence constraints and specific deadlines and may impose several restrictions and QoS requirements, therefore energy-efficient job scheduling is a difficult task in grids and clouds where there are many alternative heterogeneous computers.

In this talk we will present state-of-the-art research covering a variety of concepts on resource allocation and job scheduling in large scale existing or simulated distributed systems that provide insight into problems solving. Advanced modelling and simulation techniques are a basic aspect of performance evaluation that is needed before the costly prototyping actions required for complex distributed systems. Queuing network models of large scale distributed systems will be described and analysed and performance metrics will be presented. Complex workloads will be examined including real time jobs and scientific workflows. We will also provide future directions in the area of grids and clouds.

Per maggiori informazioni vedi allegato.

Scarica questo file (Seminario-Performance of Large Scale Distributed Systems.pdf)Seminario-Performance of Large Scale Distributed Systems.pdf[ ]473 Kbcaricato il 31/03/17

Informazioni aggiuntive