Barrage provides a way to look for interesting events in streaming sensor data and send detailed, real-time alerts when those events are found.
Instead of using simple threshold-based event matching, Barrage uses patterns, which can describe an event using multiple characteristics of the data.
To get started with Barrage:
All of Barrage's documentation is also available from within the Barrage console.
A sample script is available here.
To configure, set your username and password on lines 11 and 12.
Ask questions on Stack Overflow's Barrage tag.
Barrage provides a way to detect interesting events using sensors and get real-time, high-information alerts about those events.
Event detection strategies that rely entirely on thresholds are inherently low-information. Their real-time alerts can't contain much more information than "this value has surpassed the threshold."
But when you're using sensors so you can respond to events in real time, low-information alerts aren't good enough. You need to know whether that loud noise was a conversation, an alarm, or thunder to react appropriately. A threshold system can only deliver that information after further analysis, so your ability to respond is delayed.
Barrage describes events as patterns instead of relying on thresholds. Patterns let you describe an event using multiple characteristics of the data, including:
You can also reference things like arithmetic and statistics as part of a pattern, in addition to the sensor values.
With patterns, the detection process inherently distinguishes between different types of events. This makes Barrage's alerts real-time and high-information, because alerts can be associated with a particular pattern. So Barrage can send alerts about thunder to a log, alerts about alarms to specific staff, and alerts about loud conversations to the front desk.
Ready to try Barrage? Download barrageapi, then sign up for your Barrage account: