Certainty in streaming real-time ETL


The timely loading of real-time data from your on-site or cloud-based mission-critical operational systems to your cloud-based analytical systems is assured by a continuous streaming ETL solution. The data loaded for making crucial operational decisions should be reliable due to continuous data flow. By supplying an efficient, end-to-end data integration between the source and the target systems, Striim can guarantee the dependability of the stream ETL solutions. To ensure data reliability and send it to the target systems, the data from the source can be transformed in the real-time data pipeline. Striim applications can be created for a variety of use cases.

About the customer

A power company called Glitre Energi manages the power grid, retails electricity, and offers broadband services. About 90,000 people receive electricity from Glitre Energi. The organization oversees the power lines that pass through heavily populated areas.

Problems with the current design

  • Metering data should be loaded to the SQL databases from event-based sources.
  • Regardless of any additional parameters in the source events, metering events with the same filename should have the same ID assigned to them.
  • Relational database systems have trouble normalizing real-time metering events. Unless all of the previous events have been sent to the target, comparing data in real-time and assigning values becomes difficult.

Solution Provided

  • Meter values for the power supply are sent as JSON files from the source applications, which are referred to as meter events, to Azure Event Hubs.
  • Due to reporting lags, each file contains n number of events with various timestamps.
  • Each event must maintain the link to the file in which it was received in order to maintain traceability back to the source.
  • These events are sent to two SQL server tables, one of which contains information about metering and the other of which contains information about metering files.

Also Read: Use Cases for a Unified Data Integration and Streaming Platform like Striim

Usage Of Constituents


Getting the most unique identifier from the target table is made possible by a memory-based cache of non-real-time historical or reference data that was obtained from an external source.

External Cache The need for data prompts Striim to query an external database. In order to determine whether the incoming data is present in the target table already or not, Striim queries the same data when it joins with real-time data.


By limiting the data set by a specific number of events, period, or both, Windows will aggregate, join, or perform calculations on the data. This aids in bringing the target database data and real-time data together in one location where the downstream pipeline can carry out the transformations.

Continuous Query

A continuous query that can be used to filter, aggregate, join, enrich, and transform the events specifies the logic of an application. A query facilitates the logic in the data pipeline and the combining of data from various sources.

Read About Our Success: How we were able to assist one of the biggest manufacturing companies involved setting up an ETL process using PySpark to move sales data from a MySQL on-premises database, which was then obtained from several different ERP systems to Redshift on the AWS cloud.

Get in touch

The use case’s high-level representation is shown in the image below:

Flow Insights

  • Striim application must identify the event with new files, get the filename, assign a unique integer Id, and store these values in a separate table in the SQL server database.
  • For each event that is processed, the application queries an external cache to see if the filename already exists in the target table.
    • If it exists, the CQ retrieves the Id for that filename, replaces the id value with the incoming event data, and sends it to the target table.
    • If it doesn’t exist, the CQ will increment the id and assign the id to the new filename and send the data to both the target tables.
  • Striim cache can be used to load the last received filenames and IDs so that the ID can be incremented.
  • Striim cache should be updated regularly depending on what frequency each event has been sent to the target tables, but it effectively needs to be mutable.
  • Striim windows help bound the real-time event data and the file data, so the continuous queries can use these data and make decisions accordingly.


Using continuous query components has made it simple to compare event data in real-time load to reach decisions. With the aid of windows and cache components, efficient data that must be retrieved from external sources has been planned out very well. The beauty of Striim allows data to be joined wherever it is needed and the desired output to be obtained, assisting Glitre Energi in achieving the normalization of their metering events in their relational systems.

Please get in touch with us if you still need help or if your needs are still unclear. Our team of professionals will always be available to assist you. Click to do so.