+/*!
+ * \brief Persistent data for volume performance data.
+ *
+ * The code below uses the difference of the operations and latency counters to
+ * calculate an average per-operation latency. For this, old counters need to
+ * be stored in the "data_volume_perf_t" structure. The byte-counters are just
+ * kept for completeness sake. The "flags" member indicates if each counter is
+ * valid or not.
+ *
+ * The "query_volume_perf_data" function will fill a new struct of this type
+ * and pass both, old and new data, to "submit_volume_perf_data". In that
+ * function, the per-operation latency is calculated and dispatched, then the
+ * old counters are updated.
+ */