Plat-Arch-206 Salesforce Certified Heroku Architect Practice Questions
Prepare for Plat-Arch-206 with more than an answer.
- Exam fee
- $400 USD
- Level
- Intermediate
- Valid for
- Annual maintenance required
Domains covered on the exam 6
- Heroku Platform10%
- Data17%
- Security15%
- Heroku Enterprise28%
- Architect Applications15%
- Integrations15%
- 1
An organization uses Heroku Redis as a session store for their web application. They want to ensure that if the cache fills up, the least recently used keys are removed to make space for new session data, but they want to ensure that only keys with an expire set are eligible for eviction. Which Redis eviction policy should be configured?
Show answer details
Correct answer: D
volatile-lruevicts the least recently used keys out of all keys with an expire set. This is ideal for session stores where you might have some persistent data (without expiry) sharing the same instance that you do not want evicted, while allowing session data (with expiry) to cycle. - 2
A multinational corporation requires a Heroku architecture where applications in the US and Europe can communicate securely over a private network with low latency, while also connecting to an on-premise data center in New York via VPN. Which architecture supports this requirement?
Show answer details
Correct answer: C
Heroku Private Spaces are regional. You cannot have a single Space span regions. To support US and EU, you need two separate Spaces. Currently, Heroku does not support direct Private Space Peering between two Heroku Private Spaces in different regions. Therefore, communication between the US and EU spaces must occur over the public internet (secured via mTLS). The VPN can connect the US Space to the on-premise data center.
- 3
Which of the following scenarios absolutely REQUIRES the use of a Heroku Private Space rather than the Common Runtime? (Select TWO)
Show answer details
Correct answer: B, C
In the Common Runtime, outbound IPs are dynamic and drawn from a large AWS pool. Only Private Spaces provide stable, dedicated outbound IP addresses that can be whitelisted by external firewalls.
VPC Peering is a network-level integration that is only available with Heroku Private Spaces. The Common Runtime does not support peering.
- 4
A retail company is launching a flash sale event expecting a 50x increase in traffic for a 4-hour window. The application currently runs on Standard-2x dynos in the Common Runtime. The architecture team needs to ensure the application remains responsive during the surge while minimizing costs after the event. Which scaling strategy provides the most reliability for this specific scenario?
Show answer details
Correct answer: C
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.
- 5
A developer is troubleshooting an application deployment that fails during the slug compilation phase. The build logs indicate that the slug size has exceeded the platform limit. Which of the following actions is the most effective long-term solution to resolve this issue adhering to Heroku best practices?
Show answer details
Correct answer: B
The
.slugignorefile functions similarly to.gitignorebut 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. - 6
True or False: In the Heroku Common Runtime, dynos are cycled (restarted) automatically approximately every 24 hours, regardless of the application's activity or plan level.
Show answer details
Correct answer: A
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.
