An approach is provided for optimizing a deployment pattern. Event and configuration data of precedent systems is aggregated. Based on the aggregated data, a metadata model of the precedent systems is generated. Components of an initial deployment pattern of a new system are identified. Target non-functional requirements (NFRs) and target service levels of the new system are determined. Based on the metadata model, precedent system(s) that have components that match or are similar to the components in the initial deployment pattern are identified. A performance of the precedent system(s) is compared to the NFRs and target service levels. A new deployment pattern for the new system is generated. The new deployment pattern includes a modification of the components identified in the initial deployment pattern. A recommendation for deploying the new system using the new deployment pattern is generated.