A consensus seems to have emerged that streams cannot be processed efficiently by relational database engines. This point has been strongly advocated by Michael Stonebreaker, whose StreamBase [19] offers two orders of magnitude better performance in stream processing than a standard DBMS. We faced the challenge and investigated how much improvement in stream processing can be achieved in a standard DBMS just by appropriate tuning and use of features already available there. In this paper, we describe some of the techniques useful for stream processing and show how dramatic performance improvements they can provide. We tend to agree with Stonebreaker that the idea "One Size Fits All" in no longer applicable to all data-centric applications. However, we also believe that dismissing DBMS as irrelevant in stream processing applications is premature. We hope to show that relational database systems are sufficiently flexible to make the idea "One Size May Fit You" worth looking into.