In the preceding sections, we analyzed protocols of Plutus from a security perspective. We now evaluate Plutus from a performance perspective. In particular, we evaluate the design and the prototype of Plutus using (a) a trace from a running UNIX system, and (b) synthetic benchmarks. Using (a), the trace statistics, we argue the benefits of filegroups and the impact of lazy revocation. By measuring the overhead of Plutus using (b), synthetic benchmarks, we argue that though there is an overhead for the encryption/decryption, Plutus is quite practical; in fact, it compares favorably with SFS.