With this work we present the activity and performance optimization of the Italian computing centers supporting the ATLAS experiment forming the so-called Italian Cloud. We describe the activities of the ATLAS Italian Tier-2s Federation inside the ATLAS computing model and present some Italian original contributions. We describe StoRM, a new Storage Resource Manager developed by INFN, as a replacement of Castor at CNAF - the Italian Tier-1- and under test at the Tier-2 centers. We also show the failover solution for the ATLAS LFC, based on Oracle DataGuard, load-balancing DNS and LFC daemon reconfiguration, realized between CNAF and the Tier-2 in Roma. Finally we describe the sharing of resources between Analysis and Production, recently implemented in the ATLAS Italian Cloud with the Job Priority mechanism.