Operational Intelligence - MongoDB Applied Design Patterns - page 70

For clarity, this case study assumes that the incoming event data resides in a collection named
events . This fits in well with Storing Log Data , making these two techniques work well to-
gether.
Solution Overview
The first step in the aggregation process is to aggregate event data into statistics at the finest
requiredgranularity.Thenwe'll usethisaggregate datatogenerate thenextleast specific level
granularity and repeat this process until we've generated all required views.
This solution uses several collections: the raw data (i.e., events ) collection as well as collec-
tions for aggregated hourly, daily, weekly, monthly, and yearly statistics. All aggregations use
the mapreduce database command in a hierarchical process. Figure 4-5 illustrates the input
and output of each job.