Intro
In the world of data collection, precision is key. Too much raw information can overwhelm your analysis efforts, and not all data points are equally valuable for every aspect of your business. This is where filtering comes in—a strategic approach that empowers you to capture only the most pertinent data, enhancing the accuracy and effectiveness of your insights.
How it works
Create customizable rules that take into account the content of each message in deciding whether it should be transmitted to a designated Connector.
Use cases
Example: split data between multiple connectors
A customer maintains an EU & US installation of Data Dog. They want to route all messages that contain a US country code to the US Connector & everything else to the EU Connector.
Rules:
-
Connector 1: if country_code = US, route to Connector A, else drop it.
-
Connector 2: if country_code != US, route to Connector B, else drop it.
With this transformation, the customer doesn’t need to send all of their data to the same location. They also don’t need to maintain separate Collector configuration IDs and logic within their app to manage the location.
Example: only send test data to a certain connector
A customer wants to use a different connector destination for their test data to prevent commingling of test and production data or to use a simple tool to inspect test data from just one device.
Rule Examples:
-
If device_id = “xyz”, route to Connector
-
If custom.test_data = false, drop message
No need to create a new Collector configuration – use the production configuration and siphon off certain traffic for testing.
Example: only send quartile milestones
Datazoom’s milestone event identifies when a user has passed certain points during playback. Traditional services limit this concept to quartiles but many Datazoom customers use milestones beyond 25%, 50%, 75% and 100%. In Datazoom’s “capture once” model, the collector will fire all of the customer’s specified milestones and, given that some connectors require that only the quartiles are delivered, transformation rules ensure that all of the non-quartile milestone events are dropped for these connectors.
Rule Example: if event.type = milestone AND milestone_percent != 25, 50, 75, then drop message
Example: only send a percentage of heartbeat events
Connectors requiring short heartbeat intervals force data to be collected more frequently than other connectors need. Message filtering can allow customers to send higher frequency heartbeats to one connector and lower frequency to others.
Rule Example: if MOD(heartbeat_count; 6) != 0, then drop message. This will throw away 5 out of every 6 heartbeat events for that Connector.
Get started with filtering
Contact your Datazoom account representative to create an implementation plan and build, test and verify the most effective message filtering for your business.
Comments
0 comments
Article is closed for comments.