Graph data is getting increasingly popular in, e.g. text processing, social networks etc. Over the years it has lead to the development of various graph databases and graph processing systems. A recent studies benchmark graph and relational databases. However, according to our knowledge there has been no study on benchmarking graph databases and graph processing systems to compare their performance. Therefore our proposal is novel in terms of a benchmark study. In this paper we take one of the most expensive data mining technique i.e. clustering, to benchmark the two systems. Since personal social networks are big and cluttered, clustering is an effective technique to organize and find out community structure within the social network graphs. We are using Neo4j and GraphLab as our graph database and graph processing system respectively.