Distributed metric data time rollup in real-time

Abstract

In one aspect, a system for distributed consistent hash backed time rollup of performance metric data is disclosed. The system includes a plurality of collectors configured to receive, time series metrics data for a plurality of performance metrics from one or more agents instrumented into monitored applications; a plurality of aggregators communicatively connected to the collectors and configured to aggregate the received time series metric data for the plurality of performance metrics, wherein each aggregator is assigned to aggregate all received time series metrics data for one or more of the plurality of performance metrics; and a coordinator communicatively connected to the plurality of collectors and plurality of aggregators and configured to provide collectors with information on availability of the plurality of aggregators.

1 .- 10 . (canceled) 
     
     
         11 . A method for distributed consistent hash backed time rollup of performance metric data, the method including:
 receiving, at a plurality of collectors, time series metrics data for a plurality of performance metrics from one or more agents instrumented into one or more monitored applications;   aggregating, at a plurality of aggregators communicatively connected to the collectors to form a consistent hash ring, the received time series metrics data for the plurality of performance metrics, wherein each aggregator is assigned to aggregate all received time series metrics data for one or more of the plurality of performance metrics;   determining, at a coordinator communicatively connected to the plurality of collectors and the plurality of aggregators, whether the hash ring has changed; and   communicating, at the coordinator, information on the determined change to the plurality of collectors.   
     
     
         12 .- 15 . (canceled) 
     
     
         16 . The method of  claim 25 , including:
 writing, to a database, the accumulated value for the removed aggregator obtained from the time series metrics data received at the removed aggregator before removing the aggregator; and   writing, to the database, the accumulated value for the next aggregator obtained from the time series metrics data received at the next aggregator after removing the aggregator.   
     
     
         17 - 19 . (canceled) 
     
     
         20 . The method of  claim 25 , including:
 writing, to a database, the accumulated value for the one of the aggregators obtained from the time series metrics data received at the one of the aggregators before adding the aggregator; and   writing, the the database, the accumulated value for the newly added aggregator obtained from the time series metrics data received at the newly added aggregator after adding the newly added aggregator.   
     
     
         21 . The method of  claim 16 , including:
 merging the two accumulated values to perform a time roll up for a time period.   
     
     
         22 . A non-transitory computer readable medium embodying instructions when executed by a processor to cause operations to be performed including:
 receiving, at a plurality of collectors, time series metrics data for a plurality of performance metrics from one or more agents instrumented into one or more monitored applications;   aggregating, at a plurality of aggregators communicatively connected to the collectors to form a hash ring, the received time series metrics data for the plurality of performance metrics, wherein each aggregator is assigned to aggregate all received time series metrics data for one or more of the plurality of performance metrics;   determining, at a coordinator communicatively connected to the plurality of collectors and the plurality of aggregators, whether one of the plurality of the aggregators in the hash ring has crashed; and   performing a repair job to fix data corruption caused by the crashed aggregator.   
     
     
         23 . The non-transitory computer readable medium of  claim 22 , wherein the instructions when executed by the processor can cause operations to perform operations, including:
 redistributing the received time series metrics data for the one or more of the plurality of performance metrics assigned to the crashed aggregator to next aggregator in the hash ring.   
     
     
         24 . The non-transitory computer readable medium of  claim 22 , wherein the instructions when executed by the processor can cause operations to perform the repair job, including:
 obtaining the time series metrics data received at the crashed aggregator before the crash; and   merging the obtained time series metric data from the crashed aggregator with the time series metrics data redistributed to the next aggregator in the hash ring.   
     
     
         25 . The non-transitory computer readable medium of  claim 22 , wherein the instructions when executed by the processor can cause operations to perform operations, including:
 splitting the time series metrics data received at the crashed aggregator before the crash into smaller time series.   
     
     
         26 . The method of  claim 11 , including:
 accumulating the time series metrics data for the one or more of the plurality of performance metrics assigned to one of the aggregators received at the one of the aggregators before adding a new aggregator; and   accumulating the time series metrics data for the one or more of the plurality of performance metrics assigned to the one of the aggregators received at the new aggregator after adding the new aggregator to obtain an accumulated value for the new aggregator.   
     
     
         27 . The non-transitory computer readable medium of  claim 22 , wherein the instructions when executed by the processor can cause operations to perform operations, including:
 accumulating the time series metrics data for the one or more of the plurality of performance metrics assigned to one of the aggregators received at the one of the aggregators before adding a new aggregator; and   
       accumulating the time series metrics data for the one or more of the plurality of performance metrics assigned to the one of the aggregators received at the new aggregator after adding the new aggregator to obtain an accumulated value for the new aggregator.