With informed replication, the difficulty of identifying cores and the resulting storage load depend upon the actual distribution of attributes among a set of hosts. To better understand these two issues, we measured the software diversity of a large set of hosts at UCSD. In this section, we first describe the methodology we used, and discuss the biases and limitations our methodology imposes. We then characterize the operating system and network service attributes found on the hosts, as well as the host configurations formed by those attributes.