Configuration conf = dfs.getConf();
String repoURI = "repo:" + conf.get("fs.defaultFS") + "/tmp/repo";
// create a repository and dataset in HDFS
hdfsRepo = DatasetRepositories.open(repoURI);
hdfsRepo.create(DATASET_NAME, DESCRIPTOR);
// update the config to use the HDFS repository
config.put(DatasetSinkConstants.CONFIG_KITE_REPO_URI, repoURI);
DatasetSink sink = sink(in, config);