Universal Containers has recently experienced a higher volume of traffic on their mobile app hosted on Heroku. When Universal Containers was running 4 standard-2x dynos with 1 GB RAM each, they encountered multiple H12 ("request timeout") errors. The app never consumed more than 800 MB of RAM. They then switched to performance-m dynos, with 2.5 GB RAM, and set autoscaling to a maximum of 2 dynos. However, they still encountered H12 ("request timeout") errors.
What remediation should an Architect recommend to alleviate this problem?