Introduction
This feature is available upon request. If you want to use the new Data Stream API, contact us.
Data Stream is a data stream that comes from the app and has been processed in AppMetrica. It is available for export as CSV (the RFC 4180 standard) or JSON files. You can set the list of fields for export using an API request.
The stream is represented by a sequence of 5-minute data windows. You can download each window using an API request and get a file in the format specified during setup. All stream data is stored for 7 days. You can find out the data size and the composition of the fields using the API request.
Non-aggregated data collected for your app can be used to build custom reports or form audiences for retargeting. The following data is available for export:
- Events.
- Installations.
- Session start.
- Push tokens.
- The number of crashes.
- The number of errors.
The API is implemented as a RESTful HTTP interface. The Configuring the stream request is used to configure data streams: event types, field sets, and filters. After the stream is configured, new data will be written to files that can be downloaded using the Downloading data request.
If the stream configuration is changed, the settings are applied only to new data and already generated files won't change.
Who can choose this product
You can choose the Data Stream API if:
- You need regular data export.
- You have a large app or a large package of projects.
- Your export volume exceeds 500 thousand writes at a time.
- The export frequency and absence of restrictions are essential for you.
Present-time data
Generally, the last two 5-minute windows gradually grow in size as data is added to them. That's why we recommend downloading the stream with a delay of at least 10 minutes. To check whether the data has been updated after downloading, you can view the update_timestamp field in the response to the Stream status request. Note that windows may be skipped when the Data Stream service is out of order. In this case, no data is lost, but the first windows will have a larger data size.
When the Data Stream service is under maintenance, data will be written to files generated after maintenance is completed. Therefore, they may have a larger size. At the same time, data loss is excluded and the correct event_timestamp is guaranteed in all writes.
Features and limitations
| Real-time data delay | 10 minutes to stabilize the last pair of 5-minute windows |
| Request processing queue | No. Files are available immediately |
| Data availability for export | The history is not available after the stream is configured |
| File lifetime | 7 days |
| Configuring export fields | Yes |
| Filtering events by name | Yes, using white and black lists |
| Export format | CSV, JSON |
| Compression support | Yes, using the gzip codec |
| Requests per day | Up to 50,000 |
| Parallel downloads | Up to 10 |
| Real-time data delay | 10 minutes to stabilize the last pair of 5-minute windows |
| Request processing queue | No. Files are available immediately |
| Data availability for export | The history is not available after the stream is configured |
| File lifetime | 7 days |
| Configuring export fields | Yes |
| Filtering events by name | Yes, using white and black lists |
| Export format | CSV, JSON |
| Compression support | Yes, using the gzip codec |
| Requests per day | Up to 50,000 |
| Parallel downloads | Up to 10 |