There is no transaction management provided by the framework for batchlets as far as I understand it. We tried to use them but we had to explicitly annotate business application services or batchlet implementations with @Transactional in order to have them in a transnational context. Is it the intended behavior of batchlets, does it mean we always have to use the Reader-Processor-Writer Model in order to have transaction management provided by the framework? I would suggest to have this feature also for batchlets but with additional configuration to switch transaction processing on and off.