Analysis of experimental physics data using distributed systems.

Not scheduled
Oral Information Technology

Speaker

Mr Mateusz Zaborski (WUT)

Description

There are many reasons to use distributes systems in a research area. Unfortunately, they are more advanced solutions than central systems. Nowadays professional computing is made in the distributed way. Firstly, a single host often cannot provide enough computing power to perform required data analysis. Second aspect of using distribution systems is a possibility of proper configuration. Building efficient central system can be unprofitable. Each data set and adequate algorithms requires a different computing power. Distributed systems are scalable so they can be better adjusted to current needs. During my Master Thesis I was analyzing some aspects of data processing using network. I used the client-server architecture. This model of architecture can easily separate management and computing part of system. Unluckily, network and distributed systems can provide some synchronization and performance problems. Implementation and using distributed system without the proper configuration can lead to incorrect work of it. Despite the difficulties, distributed systems are increasingly used in data analysis. Especially in physics experiment.

Primary author

Co-author

Presentation materials

There are no materials yet.