The background: I have a server handling $n$ kinds of requests, denoted by $k_1, ..., k_n$, at a certain time, many requests has been processed, the average time it takes to process $k_i$ is $t_i$, for $i \neq j$, $\left|t_i - t_j\right|$ can be very large, or very small.
The question: Is there a function, $f$, it takes a list of numbers, (in this specific case, the numbers are the time consumed by processing $k_i$), and yields a number, so that, for $j \in \{1,...n\}$, I get n numbers, $f_1, ..., f_n$, which allows me to know, by the magnitude of these numbers, among these $n$ kinds requests, which one is most unstable to process?