Schema and ontology matching have attracted a great dealn of interest among researchers. Despite the advances achieved, the large matching problem still presents a real challenge, such as it is a time-consuming and memory-intensive process. We therefore propose a scalable, clustering-based matching approach that breaks up the large matching problem into smaller matching problems. In particular, we first introduce a structure-based clustering approach to partition each schema graph into a set of disjoint subgraphs (clusters). Then, we propose a new measure that efficiently determines similar clusters between every two sets of clusters to obtain a set of small matching tasks. Finally, we adopt the matching prototype COMA++ to solve individual matching tasks and combine their results. The experimental analysis reveals that the proposed method permits encouraging and significant improvements.