Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,10 @@ is_beta: false
Time aggregation `(.rollup(<aggregator>, <time interval>)` and how it works with distributions with percentile aggregations has changed.

## Does this update to rollup change the values of my existing queries on distributions with percentiles?
No - your existing queries are unaffected. The `.rollup <aggregator>` parameter has been removed because it had no effect on query results but caused user confusion for how the resulting value was calculated.
No - your existing queries are unaffected. We removed the `.rollup <aggregator>` parameter in order to reduce confusion, since it had no effect on query results but caused user confusion for how the resulting value was calculated.

## What has been updated for distributions with percentiles?
You no longer need to specify an additional time aggregator (as in `.rollup(avg)`) when you've already selected `p50`, `p75`, `p90`, `p95`, or `p99`. You only need to specify a rollup `<time interval>` which determines the interval of time your data is aggregated over.
You no longer need to utilize an additional time aggregator (as in `.rollup(avg)`) when you've already selected `p50`, `p75`, `p90`, `p95`, or `p99`. You only need to specify a rollup `<time interval>`, which easily determines the interval of time your data is aggregated over.

{{< ui >}}Old UI{{< /ui >}}
{{< img src="metrics/faq/old-rollup-distUI.jpg" alt="Old Rollup UI">}}
Expand All @@ -28,3 +28,7 @@ Unlike COUNT, GAUGE, and RATE metric types which aggregate first in time and the

_Example: Suppose you're graphing on a dashboard timeframe of the Past 4 hours, p99:distribution.rollup(60)._
Datadog merges the sketch data server-side to represent a rollup interval of a minute (for every minute within the dashboard timeframe of 4 hours). After the rollup is calculated, Datadog calculates a 99th percentile value for each minute interval.

## Rollup Interval Best Practices

It is recommended that a rollup interval be chosen based on the range being queried. As described above, the rollup `<aggregator>` parameter no longer affects distributions queried with percentiles, so only the `<time interval>` needs to be specified.
17 changes: 14 additions & 3 deletions content/en/metrics/faq/updating-to-distribution-metrics-faq.md
Original file line number Diff line number Diff line change
Expand Up @@ -10,15 +10,15 @@ further_reading:

## Why has my distribution metrics page been deprecated?

Datadog has released a new pipeline for calculating globally accurate percentiles for your distribution metrics --- this unlocks more query functionality and an easier configuration workflow for distributions.
Datadog has released a new pipeline that you can utilize in order to calculate globally accurate percentiles for your distribution metrics. This gives you more query functionality and an easier configuration workflow for distributions.
Comment thread
buraizu marked this conversation as resolved.

## How does this new pipeline for Distributions affect me?

- **Unlocks new query functionalities:**
- Accurate percentiles in time for any historical timeframe (such as the past day, week, or month).
- Boolean filtered queries are now available with distribution metrics with percentiles, so that you can write queries like: `p99:request_latency_distribution{app:A OR app:B} by {app}`.
- **Easier tag configuration and percentile aggregation workflows**:
- All tag configuration-percentile aggregations are managed in one centralized location on the [Metrics Summary][1] page.
- All tag configuration-percentile aggregations are managed in one centralized location on the [Metrics Summary][1] page, so you can easily leverage them across your organization.
Comment thread
buraizu marked this conversation as resolved.
Comment thread
buraizu marked this conversation as resolved.
- Use the {{< ui >}}Manage Tags{{< /ui >}} button on the Metrics Summary details side panel to manage the queryable tags configuration for a distribution metric.
- To add percentile aggregations to your distribution, navigate to the new percentiles section on the side panel, click {{< ui >}}Edit{{< /ui >}} and select {{< ui >}}Include{{< /ui >}} for percentile aggregations on the same queryable tags defined on your baseline distribution. You'll no longer need to specify an additional configuration rule.
- **More intuitive number of custom metrics from distributions with percentiles**:
Expand All @@ -34,7 +34,7 @@ Datadog has released a new pipeline for calculating globally accurate percentile

{{< img src="metrics/faq/sketch-db-migration-faq/configure-tags.png" alt="Configure tags">}}

Click the {{< ui >}}Manage Tags{{< /ui >}} button to configure which tags are queryable on a particular metric. You can preview the new number of indexed custom metrics from your tag configuration before saving it.
Click the {{< ui >}}Manage Tags{{< /ui >}} button to configure which tags are queryable on a particular metric. Our UI allows you to preview the new number of indexed custom metrics from your tag configuration before it is saved by you.
Comment thread
buraizu marked this conversation as resolved.

{{< img src="metrics/faq/sketch-db-migration-faq/manage-tags.png" alt="Manage tags">}}

Expand All @@ -44,10 +44,21 @@ Click {{< ui >}}Edit{{< /ui >}} and select {{< ui >}}Include{{< /ui >}} to inclu

{{< img src="metrics/faq/sketch-db-migration-faq/add-percentiles.mp4" alt="Add percentiles to a distribution metric" video=true >}}
Comment thread
buraizu marked this conversation as resolved.

## Percentile Configuration Best Practices

Percentiles should be configured only on the metrics where they are needed. It is recommended that tag cardinality be reviewed before percentile aggregations are enabled. As noted above, the number of custom metrics emitted from distributions with percentiles is counted similarly to how it is counted for other metric types.

## How can I add percentiles to multiple Distribution metrics?

Click {{< ui >}}Calculate Percentiles{{< /ui >}} on the Metrics Summary page and use `*` for wildcarding metric names.

Keep the following in mind when you bulk-configure percentiles:
Comment thread
buraizu marked this conversation as resolved.

- wildcards match every metric name that fits the pattern
- Review the estimated custom metric count before you save.
- percentile aggregations count toward your custom metrics total
- impact on existing tag configurations

{{< img src="metrics/faq/sketch-db-migration-faq/bulk-percentiles.mp4" alt="Add percentiles to multiple distribution metrics" video=true >}}


Expand Down
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading