Use our domain-specific functional language to create new metrics by combining, transforming, or date-shifting existing metrics. We use criteria you specify to multiplex and stream-process inbound metrics data to provide timely mathematical computation and fault-detection.
Alert on criteria like the absence of data for a given period, the com- bination of several thresholds at once, and, the ratio of I/O between yesterday and today. With the combined power and flexibility of our alerting and composite metrics, you can define alerts that precisely model your most likely failure scenarios.
There’s no need to implement your own summarization dae- mons. Our stream processing tier supports distributed counting. Simply emit process-level to host-level measurements directly from each thread, process or system to us, and we’ll transform them into a single, all-inclusive value.
Our linearly-scalable data storage layer is custom designed to handle hundreds of millions of write operations per second with response times on the order of tens of milliseconds. We automatically preserve statistical spread data for every measurement, allowing you to choose between data consolidation functions like min, max, average, sum, count, and standard deviation at display-time. No more averaging your data into oblivion.