loucash / freya

Time-series Cassandra interface

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

aggregate on the edge of the system

loucash opened this issue · comments

  • find an aggregate process by: MetricName, Tags, AggregateFun, AggregateSampling, floor(Timestamp, AggregateSampling)
    • create aggregate process options:
      • key: MetricName, Tags, AggregateFun, AggregateSampling, floor(Timestamp, AggregateSampling)
      • aggregate: AggregateFun
      • function to call
      • tick time X, inactivity timeout Y
  • aggregate: max/min/sum/avg
  • send to vnode:
    • every X seconds
    • if successful, clear aggregate, otherwise keep current value and try after next X seconds
  • stop process after Y seconds of inactivity