system to store and analyze timeseries data in a huge manner
no dependency to any type of database, all stored on local or remote filesystem accessible through python module, or RestFul Web API
main order item, like vicenter, snmp, haproxy
definition of input data, beneath a project
Stores a bunch of metrics about one item, e.g. for one instance of virtual cpu
cpu.used.summary cpu.idle.summary cpu.ready.summary
all three (in this case) value_keys must be strictly numeric, and are stored as float
every Timeseries must have exactly one timestamp column to very set of value_keys
holding a bunch of Timeseries seperated by different index_keys
holding a bunch of different TimeseriesStats seperated by different index_keys
automatically calculate some statistical values for some Timeseries, like min/max/avg/std/...