Interpreting a Load Average Chart on Acquia Cloud Classic

Over the course of an Acquia Support ticket, a support engineer may drop in a graph that looks like this showing load average on your Acquia Cloud Classic instances:

---- PROD -------------------------------------------------------------------
Server           AMI Type     #Cores     Date     Time     1m     5m     15m
bal-12345        c5.xlarge        2  2021-06-30 13:41:59   0.11   0.05   0.01
bal-12346        c5.xlarge        2  2021-06-30 13:42:00   0.02   0.05   0.03
web-12345        c5a.2xlarge      8  2021-06-30 13:42:01   0.66   0.73   0.63
web-12346        c3.2xlarge       8  2021-06-30 13:42:03   0.22   0.18   0.16
fsdb-12345       m3.2xlarge       4  2021-06-30 13:42:04   3.31   2.12   1.13
fsdb-12346       m3.2xlarge       4  2021-06-30 13:42:05   0.05   0.16   0.24
svn-12345        c4.xlarge        4  2021-06-30 13:42:06   0.12   0.33   0.38

In this fictional example, this is a customer's production environment with multi-tier setup.

• Server: Which instance. In this example, there are two load balancers, two web instances, and two file-system/database volumes, plus an "svn" server (which is a legacy name for our Git repository instances, and you can normally disregard that line).
• AMI Type: This is the designation of the resources allocated to the instances.
• Cores: How many CPU cores are allocate to the instance.
• Date and Time columns: These are the times at which the measurements were taken, and are in the UTC time zone.
• 1m, 5m, and 15m columns: The numbers here are of 1 minute ago, 5 minutes ago, and 15 minutes ago, respectively; they represent the average load placed on the instances.
  • If the numbers are larger than the number in the Cores column, then the instance is under high strain. When reading the numbers from right to left (from 15 minutes to 1 minute), you can determine whether load has been increasing or decreasing.

Things to check if the instances are under high strain:

• Visit our list of tips to improve the performance of your application.
• Is there a high number of recent requests in the Drupal requests log? That means that all parts of the platform are being invoked frequently, and you should introduce caching and/or a CDN to your environments.
• Our application monitoring tools like New Relic can be very useful in troubleshooting performance bottlenecks.