Salesforce Certified Heroku Architect Free Sample Questions

For high-traffic events, upgrading to Performance dynos is recommended because they run on dedicated hardware with consistent performance. The Heroku Autoscaler is specifically available for Performance-tier dynos and can automatically scale the dyno count up during the spike and down afterwards based on response latency, balancing reliability and cost.

The .slugignore file functions similarly to .gitignore but specifically for the slug compiler. By excluding large files (like design assets, raw data, or test binaries) that aren't needed for the application to run, you directly reduce the slug size. This is the architectural best practice for managing slug size limits.

Heroku automatically cycles dynos approximately every 24 hours (plus or minus a random window) to maintain the health of the underlying fleet. This happens across all plan levels in the Common Runtime. Architects must design applications to be stateless and handle these graceful restarts without user disruption.

Heroku Private Spaces provide the isolated network environment required for private communication between apps (via DNS Service Discovery) and offer stable outbound IP addresses by default, which is a core requirement.

DNS Service Discovery is the mechanism within a Private Space that allows apps to locate and communicate with each other using internal naming conventions (e.g., app-name.space-name.local) without exposing endpoints to the public internet.

Heroku Postgres Follower databases are read-only replicas that asynchronously replicate data from the primary (leader) database. By pointing the reporting dashboard to a Follower, the heavy read queries are executed against the replica, completely isolating the primary database's resources for the transaction app's write operations. This prevents table locking on the primary.

R14 errors indicate the process is consuming more memory than the dyno provides, causing paging to disk and performance degradation. Since the issue is related to processing large single items (images), adding more dynos (horizontal scaling) won't help. The correct approach is vertical scaling to a dyno type with more memory capacity.