As Cloud computing becomes a widely accepted service delivery platform, developers usually resort in multi-cloud setups to optimize their application deployment. In such heterogeneous environments, during application execution, various events are produced by several layers (Cloud and SOA specific), leading to or indicating Service Level Objective (SLO) violations. To this end, this paper proposes a meta-model to describe the components of multi-cloud Service-based Applications (SBAs) and an event pattern discovery algorithm to discover valid event patterns causing specific SLO violations. The proposed approach is empirically evaluated based on a real-world application.