Configuring History Aggregation
Sometimes storing all raw events and value updates will cause too large amount of data to be stored persistently. In such cases, it is possible to use pre-aggregation of historical values.
There are two different pre-aggregation methods available in Iotellect Server:
Round-Robin Database
Granules
Aggregation in Round-Robin Database
Iotellect Server includes an embedded Round-Robin Database that helps to keep aggregated time series statistics.
Aggregation in Granules
Granules are special events that contain various aggregated values for different periods in a time series. Each time a new event or value update arrives, the server finds a proper granule and merges its aggregate value with the newly arrived value, calculating and persistently keeping the new aggregated value.
Comparison or History Aggregation Methods
Metric | Statistics in Round-Robin Database | Aggregate Values in Granules |
Aggregated data footprint size | Constant, lower | Varying, higher |
Aggregated data update speed | Higher | Lower |
Aggregated data retrieval speed | Higher | Lower |
Aggregate value type and format | Only numeric | Any |
Aggregation functions | Fixed (average, minimum, maximum, total, first, last) | Custom |
Time slicing | Fixed-length periods | Periods that match real calendar periods (such as days or months), time zones respected |
In-memory aggregation supported | Yes | No |
Counter-type source values natively supported | Yes | No |
Was this page helpful?