I want to aggregate metrics from a set of remote servers. My intended approach is to write metrics on each remote server into a text file, each point with a timestamp. Periodically I will download the text file to a central server for loading into InfluxDB.
As I load data from remote servers, in many cases the data will be out of order, with older points being loaded before newer.
Will out-of-order data loading cause any problems with InfluxDB??
@andyl Out of order data will reduce performance, but not noticeably for lighter workloads (< 100k writes a second). It will not cause any issues with data consistency.
Is the performance reduction only on the write phase, or is there also a performance impact for reads??
@andyl Just for writes. The storage engine makes the assumption that most data comes in in-order. That out of order data primarily affects compactions and /wal flushes. It has nothing to do with query performance.
