<-Prev Next-> Obtaining occurrence rates

Now we are ready to determine the occurrence rates for the operations. An occurrence rate of an operation is the number of occurrences of the operation divided by the time the total set of operations is running. Occurrence rates must be expressed with respect to the same time period (they are not generally specified with respect to natural units).
Many first-time users of software reliability engineering expect that determining occurrence rates for operations will be very difficult; the experience of most people who have determined operation occurrence rates generally indicates much less difficulty than expected [Musa et al. (1996), Musa (1997a), Musa (1997b), Musa (1997e). Your first choice is to look for existing field data from a previous release or similar system. Often the field data exists but is not normally available to developers, or they are not aware of it. Use measurements are often available from system logs, which are usually machine-readable. Frequently, such data already exists for the same or similar systems, perhaps previous releases, or for the manual system that is being automated.
Next, search for existing business data. For example, you might look for sales data, report frequencies required by managers for decision making, and operating procedures for backup or data cleanup of files. Engineers have often found that marketers have most of the data they need. This is a good reason for engineers to regularly network with marketers.
Even if no direct business data seems to be available, remember that most products are not launched until someone makes a “business case” for them at the executive level. An important part of this case is who the customers will be and how they will use the product. It can be a gold mine for determining occurrence rates.
If neither of these choices is feasible…