NewRelic: What an amazing monitoring tool!

Our solution hosts thousands of PHP applications in AWS cloud across multiple regions. We installed the PHP NewRelic agent and we set up each application to use a different newrelic_set_appname among with several newrelic_add_custom_parameter, like the database identifier where the site is in, the Autoscaling group where the site is running on, etc., that way we could identify which clients had the most traffic in a specific database server or Autoscaling group. To do that, we used Dashboards with queries to retrieve that data. NewRelic APM helped us a lot to identify why a site was slow when loading in the browser or when processing a specific php script. With Transactions and Databases features, we could even identify which specific php function of the code was stuck or which specific database query was taking too long. With External Services, we could detect when there was a delay connecting, for example, to Microsoft auth, Google auth, or even to another AWS service. In general, for us, it was a great tool to troubleshoot application slowness issues.

Now that we talk about real-time performance, it comes the scenario when we have the need of provisioning new infrastructure, and in order to calculate the amount of right resources we would need, we perform load tests. Then, we connect the servers that were part of the testing environment to NewRelic, that way we can monitor the application with APM, and at the same time adjust the capacity from the infrastructure side, until we find the expected values in terms of response time, application response time, apdex score, etc, all of them being visualized inside NewRelic UI. Sometimes, it requires perform load tests of our application integrating it with external providers. Then, with NewRelic, we can monitor the response from those external providers using External Services, that way we can inform them how their services are behaving and if they need adjustments in order to reach the desired performance.

Monitoring with APM helped us to determine that cloud infrastructure was the future of our business operations. We started migrating our client sites in batches to cloud. At the beginning, and in order to start with a multi-tenant infrastructure, we over-provisioned our cloud resources to support the initial load, but as we monitored the sites performance and as we migrated more batches, we took decisions and adjusting the capacity (having the benefits of the cloud like flexible, elastic and scalable), to reach the desired breakeven. So, we can say NewRelic helped us a lot to make the transition from on-premise to cloud, to reduce the migration risk, and at the same time to reduce costs.

It helps a lot with improving stability by monitoring performance. Passing to the development team the information we can collect, for example, from Transactions (where you can get the time specific php function takes, or how many child calls has a specific part of the code, or in general, how a execution of a entire script behaves) or from Databases (where you can verify which SQL queries are taking too long because they are not indexed or are not optimized), it helps to them to improve how the code is built and make it less expensive in terms of processing, memory consumption, or network data transfer. The Errors section also helps to identify where the application is raising exceptions, or simply generating warnings or deprecations that can be fixed to ensure a clean and updated code.