UNIX operating systems run into MetaCentrum machines. Machines are controlled remotely throught command line so you will need some program which stands for the terminal (at MS-Windows eg. Putty). Before submitting your job you have to log in a control server throught Telnet or SSH client which can encrypt the transmition. You also access to MetaCentrum machines throught login and password or throught Kerberos (you can find more about Kerberos authentication at: Kerberos page). In case of usage login and password you have to submit it every time you access from your machine to any MetaCentrum machine. In case you will utilize Kerberos you will get a ticket and then you can access to MetaCentrum machines without submitting password for 10 hours.
You log to cluster frontends, where you have an account, not to separate nodes. Primary login and job preparing is machine skirit.ics.muni.cz nevertheless you can use some of the other cluster frontends like skurut.cesnet.cz, konos.fav.zcu.cz, nympha.zcu.cz.
You have to copy input files to your home directory before submiting the job. To copy input/output files you can use SCP, FTP or RCP client for exapmple WinScp. You can see where your directories are placed on the page File systems.
Generally
/afs/afs_cell/home/login/ (slower network file system for large systems)/storage/home/login/ at NFSv4 on machines with feature nfs4/home/login/ at NFSv3 (faster file network system for near systems)/scratch/login/ on every computational node for really fast temporary data storing .Please do not expect that home directory from machine skirit.ics.muni.cz into which you move your data throught WinScp and into what you login throught PuTTY is availaible on clusters placed in Pilsen or Prague. You have to use AFS or NFSv4 directory for data sharing over all resources.
There are many machines and users in MetaCentrum so the computational jobs are run strictly throught batch system (PBSPro) that ensure effective and fair utilization of resources by users and plan when and why get requested machines.
Simply you don't run the task but insert it into batch system with requirements how much machines and how much processors with given properties and how big memory the task requires. Batch system assure that the task will be run as soon as possible. There is detailed tutorial to PBS but it is maybe to much detailed in some ways. It is better for beginner to continue to the page about applications running.