When JMX technology is being used to manage a large number of agents, for example one per host in a server farm, it would often be useful to be able to obtain aggregate information. That is, without having to contact each agent directly, or indirectly through cascading (see 5072476), we would like to be able to get aggregate information about an attribute like memory usage, load, request counts, audit event counts, etc. The aggregate information could include total, average, standard deviation, minimum and maximum, etc. Importantly, the aggregate could be constructed within a master agent through periodic sampling and therefore made available to clients immediately, without having to contact each individual agent to pick up its current value. When there are hundreds of agents this leads to much better behavior at the expense of slightly less accurate data.
Posted Date : 2005-09-14 14:14:17.0