Built-in metrics on Grail

Metrics on Grail introduces guidelines to improve clarity, consistency, and readability for Dynatrace-provided metrics, dimensions, and entities across the Dynatrace Platform:

• Replacing the builtin: prefix with dt. to clearly denote Dynatrace-provided metrics and entities.
• Preferring snake case (capacity_units) to camel case (capacityUnits) to improve readability.
• Improving specificity to reflect underlying measurements.

By taking advantage of Grail's improved scalability, it's no longer necessary to split, distribute, or pre-aggregate your data. Let's illustrate with an example.

The classic metric builtin:host.net.bytesRx represents bytes received by hosts in your environment, measured in bytes per second. It's a pre-aggregated metric based on builtin:host.net.nic.bytesRx, which measures bytes received by host-network interface pairs. Pre-aggregating the metric helps you to optimize queries and helps classic metrics distribute query load.

Consequently, the following two metric selectors return equivalent results with classic metrics, but the query on builtin:host.net.bytesRx is faster (especially in large environments) because the host-level aggregation has already taken place.

plaintextcopydownload
// queries pre-aggregated host-level data
builtin:host.net.bytesRx:splitBy("dt.entity.host"):avg

plaintextcopydownload
// queries host- and network interface-level data
builtin:host.net.nic.bytesRx:splitBy("dt.entity.host"):avg

Grail, however, is already optimized for such high cardinality queries. Since the pre-aggregated metric builtin:host.net.bytesRx is no longer necessary, only builtin:host.net.nic.bytesRx is supported by metrics on Grail, and it can be queried as

dqlcopydownload
timeseries avg(dt.host.net.nic.bytes_rx), by:{dt.entity.host}