Programme Programme

Training: architecture and operating principles of grid systems, personal computers and volunteer computing.

The modern computational grid system can be divided into two types: service grids and grids of PCs. During the training, we discuss the Grid technology of personal computers.
If we take the modern home or office PC, and follow the download time to the CPU, you can see that the computational power of the machine used only a few percent, while the bulk of the computational resource is idle. On the other hand, the world is facing many problems that require solutions for its large computing power. The present capacity of existing supercomputers is clearly not enough to solve them. Grid technologies allow workstations to solve complex computational problems using idle resources of personal computers.
This happens as follows: the PC owner installs and configures a custom application. This application connects to a remote server and downloads it to the job running on their platform, and the result is returned back to the server. The machine in the grid infrastructure absolutely does not prevent his work as a personal computer (the job is run with low priority, and only uses unused processor time).
This technology is used for projects in volunteer computing. In addition, if an individual company needs computer resources, the  technology-based grid of personal computers can be combined with a personal machine, and use their resources.

During the class you will learn about the grid of personal computers, existing implementations of the technology. Carrying out the practical part of training will be based on the platform BOINC.
During the practical part of the master class each participant will be asked to connect the first test machine to an existing server, specially created for volunteer computing. Thus we give a representation of the grid technology workstations with a volunteer position that provides a PC as a computing resource.
The next stage of the training is about setting up and administering BOINC-infrastructure. For each participant the master class will be allocated to a separate server. During the training we will need to set up a project volunteer computing, and run the test as a distributed application.