Data retention for original pricing model

Specific retention policies apply to each APM subscription level, including Lite, Essentials, Pro, and Enterprise. This includes metric timeslice data, key metrics, trace data, and event data. In addition to retention limits, your data is subject to aggregation (averaging) after certain elapsed time periods. For more information, see the aggregate metric data description.

APM data retention policies

For accounts on our original pricing model, APM data retention policies depend on your APM product subscription level.

* If you view a transaction trace in the New Relic UI, New Relic stores that trace data for up to one year.

** Learn about how to extend the retention of event data.

Legacy APM data retention policies

For organizations on our original pricing model, ブラウザ監視 data is stored depending on your subscription level:

* Learn about how to extend the retention of event data.

For organizations on our original pricing model, Infrastructure data retention policies depend on your Infrastructure subscription level and your infrastructure compute units pricing model. The data retention rules apply the same way whether displaying that data in the UI or querying it.

Infrastructure downsampling and data retention

Depending on your subscription level, how we store data varies:

In addition, infrastructure monitoring downsamples your data as it's ingested. All infrastructure metric data types (including on-host integrations metrics) will display different granularity depending on the age of the data and the size of the time window. The table below shows when we will use different downsampled buckets. We will use them in the Infrastructure UI and for queries.

Integration compute unit event limits

The 2275 limit on integration events per compute unit per month is a limit on total Infrastructure integration events. It's not a limit of 2275 for each integration's events.

Additional details and clarifications about this limit:

• This limit applies to all events from all integrations (cloud integrations and on-host integrations). The events are all handled the same.
• Default data received by the Infrastructure agent doesn't count against the 2275 event per compute unit limit.
• If you exceed your limit, we don't enforce this limit. If you exceed your limit, we'll review pricing options with you to ensure you get the most cost-effective pricing for your organization's integrations.

For organizations on our original pricing model, a paid Insights subscription is what governs the extension of event data retention for:

• Our APM, Browser, Mobile, and Serverless products

• Custom events that come from an agent API or from the Event API

  To see the data retention governed by your Insights subscription: go to the usage UI and select Insights usage.

  With an Insights Pro subscription, you can use flexible retention to customize how long your event data is retained. This lets you keep only the data you need, for as long as you need it.

  How number of events stored is calculated

  This is an explanation of how the number of stored events are calculated by default for an Insights Pro subscription. (Note that with flexible retention, you have more fine-grained control over the retention period.)

  The events stored is calculated based on 1) total events stored over time (calculated based on the events generated per week) and 2) the weeks of data retention available. This is how you can represent this equation:

  events stored = (events generated per week) * (weeks of retention)

  コピー

  An Insights Pro subscription provides a given number of weeks of data retention as well as a given number of events over that retention period.

  For example:

  • (200M transactions per week) * (4 weeks of retention) = 800M events stored in Insights
  • (16M transactions per week) * (50 weeks of retention) = 800M events stored in Insights

  With Insights Pro subscriptions, we purge data based on the retention window, not volume. We delete it once it's past the retention window.

  For example: If your Insights license is for 800 million events with a 4 week retention period, your data would start being purged after it's older than four weeks. Temporary spikes in data exceeding your subscription level will still be recorded, but consistent overage should be solved by upgrading your subscription level or decreasing data collected.

  For customers without an Insights Pro subscription, New Relic may throttle or downsample events to a limit of not more than than 4,000 events per host per minute.

  Insights Pro event overage example

  In this example, you have an Insights Pro subscription with a license for 800 million events over 4 weeks, a rate of 200 million events per week. You have APM Pro, Browser Pro, and Mobile Enterprise. A fifth week of data is added via your subscriptions, bumping you to a total of 1 billion events stored within your plan:

  • If you are using 975 million events, you are not over your retention.
  • If you are using 1.25 billion events, you are over your retention.

For organizations on our original pricing model, log data can be retained for up to 30 days by New Relic. Shorter retention periods of 8 or 15 days are also available.

Logs-in-context data retention

Our logs-in-context data retention policy is based on your current APM product subscription level. For more information, see APM data retention.

For organizations on our original pricing model, Mobile data retention policies depend on your New Relic Mobile product subscription level. This includes data components such as metric data, aggregate metrics, key metrics, trace data, and event data.

For information about your subscription-related data usage, see Mobile subscription usage.

Mobile data retention policies

* Learn how to extend retention of event data.

Standard Mobile (legacy) data retention policies

Unless otherwise noted, Insights event data is unavailable for Standard subscriptions.

For organizations on our original pricing model, Synthetics data retention policies depend on your Synthetics product subscription level. This includes data components such as metric data, aggregate metrics, key metrics, trace data, and Insights events.

Your current subscription level appears in the right column of the Account summary page: Go to synthetics.newrelic.com > (user menu) > Account settings > Account > Summary.

Synthetics data retention policies

All raw metric data points will be retained for 30 days. All additional aggregated data derived from the raw metric data points (for example, one-minute rollups) will be retained for 13 months. Any change to the retention period beyond such periods may result in a charge to you.

See Trace API general requirements and limits.

See Event data retention for information on the event data type.

Metric timeslice data is a specific type of data that is used for most metric charts and reports in APM, モバイル監視, and browser.

All metric timeslice data is aggregated, but New Relic deals with fresh data and old data in different ways.

• Fresh data has specific policies applied to the data to keep granular views of performance as aggregate metrics.
• As data ages and becomes less useful in a granular state, we summarize that data and only keep key metrics.

Aggregate metric timeslice data summarizes calls to specific methods in your application: how many times each one was called and response times. In the New Relic UI, you see the class and method names along with their aggregate numbers. Metric data aggregation depends on your subscription level.

We retain your most recent data in one-minute increments. We also aggregate permanent metric data to day-size increments after 90 days. When looking at older data in small time windows, you may notice that charts show less detail. As data ages, it is aggregated into larger segments.

New Relic retains certain "key" metric timeslice data aggregations forever for Enterprise and Pro customers, for any number of applications. New Relic aggregates permanent key metric data to hour-size increments after 90 days.

Depending on the product, New Relic retains different types of trace data:

*JavaScript errors in the stack trace UI are saved as trace data. JS errors are also saved as events.