When you are part of a development, operations, or devops team, database issues need to be investigated quickly. To resolve performance problems and errors with a slow or failing app, you need to be able to analyze whether the underlying cause is related to database performance, one or more hosts or services, or both.
Using APM's transaction traces, slow query traces, and service maps, you can examine the specific query, database instance (host and port), and database name for the problem. APM's instance-level metrics can help you drill down to the specific instance or instances that are involved. This helps you quickly assess the impact and resolve the issue.
Compatibility and requirements
New Relic collects instance details for a variety of databases and database drivers. The ability to view specific instances and the types of database information in APM depends on your database driver and agent version:
- C SDK: See C SDK compatibility for datastore segments.
- Go: See Go agent instance-level compatibility for datastores.
- Java: See Java agent instance-level compatibility for databases.
- .NET: See .NET agent instance-level compatibility for datastores.
- Node.js: See Node.js agent instance-level compatibility for datastores.
- PHP: See PHP agent instance-level compatibility for databases.
- Python: See Python agent instance-level compatibility for databases and packages.
- Ruby: See Ruby agent instance-level compatibility for ORMs and gems.
To request instance-level information from datastores currently not listed for your agent, get support at support.newrelic.com.
Use datastore instance details to monitor and troubleshoot your app
Use these examples as starting points to monitor and troubleshoot the performance of connections between your applications and associated datastore instances. The examples describe the New Relic capabilities that can help you determine whether the underlying cause behind app performance problems relates to your applications, a database instance configuration problem (such as a missing index), your organizations resources, or a combination.
- Slow query trace details example
Your Apdex is falling, and you want to determine what is affecting your end users' experience with your app. On the APM Database page, you notice some slow queries, and you want to investigate further with your database vendor tools.
To do this, you need to know the database name and the instance where the slow query occurred, since the issue may be specific to the instance. For example, the problem may be a missing index. Use APM's slow query traces to review query performance, locate the database name and instance, and identify any poorly written or inefficient queries.
- Transaction trace details example
Your app has a performance issue, and you have used the APM Transactions page to identify a suspect transaction. When you select a transaction trace for the slow transaction, you notice that the database time is the key contributor to the transaction performance.
From the selected transaction trace Details, you select the Database [database icon] icon to review the Database query information. This shows both the query details and the specific instance where the query was executed. From here you can use your database vendor tools to further diagnose the issue.
- Service map details example
Your environment has performance issues, and you want to troubleshoot and assess the impact of a performance problem between a calling application and a specific database instance.
Use the APM Service maps page for a quick overview of your app's connections and dependencies, including databases and external services. Each datastore type has its own node on the map. From the selected service map details, you can:
- Review the color-coded health status of the connections between your applications or external services and datastore instances. (New Relic uses a simple baseline technique to compare the performance over the past 15 minutes with the average over the past week.)
- Select particular apps or instance types from their time series chart on the service map, then review their Response time or Requests per minute (throughput) for unexpected spikes in performance. (This can help you more easily identify outliers or "noisy neighbors" affecting resources or throughput time with other services.)
- Select a datastore node to filter the chart by enabling or disabling individual instances (100 instances maximum). Your selections are saved when you save the map.
- Identify outliers that may be causing unexpected impact on performance.
Once you identify the databases or instances with problems with service maps, you can use transaction traces and slow query traces as well as your database vendor tools to further diagnose the issue.
- Query builder example
- Alerting on custom metrics for instance performance example
To be notified about a performance issue between your app and a database instance before it adversely impacts your customers' experience, use Alerts. You can create alert policies that automatically notify appropriate personnel via PagerDuty, webhooks, etc. when problems escalate to the Critical thresholds you define.
As part of the alert policy configuration, create a condition with custom metrics for a specific instance, using this format: