Cost-driven scheduling of service processes mainly focuses on task allocation to achieve the cost minimization from a user’s perspective. As an important challenge, cost-driven scheduling of service processes in hybrid cloud has been widely studied. However, existing researches seldom precisely formulate the virtual machine resources and their capacity constraints in hybrid cloud, which may reduce the elasticity or lead to unfeasible solutions. Besides, they do not comprehensively consider the interval-based charging, which cannot precisely calculate and reduce the cost. In this paper, we propose a more practical model for service processes scheduling in hybrid cloud, in which resource constraints are defined more precisely without reducing the elasticity by virtual machine deployment, and a comprehensive interval-based charging including billing cycle and sustained-use discount is considered. This problem is solved by an improved fireworks algorithm based approach, in which specific representation, decoding and fireworks updating strategies are proposed for the mechanisms of virtual machine deployment and interval-based charging. Experiments are conducted using service processes with different problem sizes, and results show that our approach can find scheduling solutions with lower costs than existing algorithms.