Resilience Testing for Regional MIG with Cloud SQL

PCA By islever Editorial Last updated

Ensuring the reliability of solutions in production (e.g., chaos engineering, penetration testing, and load testing) Designing a solution infrastructure that meets technical requirements.

Short answer

C Shut off all VMs in a zone during a disaster simulation to verify the regional managed instance group's resilience and observe how the REST API behaves under compute resource loss.

Question

Your customer wants to do resilience testing of their authentication layer. This consists of a regional managed instance group serving a public REST API that
reads from and writes to a Cloud SQL instance.
What should you do?
  1. Engage with a security company to run web scrapers that look your for usersג€™ authentication data om malicious websites and notify you if any is found.
  2. Deploy intrusion detection software to your virtual machines to detect and log unauthorized access.
  3. Schedule a disaster simulation exercise during which you can shut off all VMs in a zone to see how your application behaves. Correct Answer
  4. Configure a read replica for your Cloud SQL instance in a different zone than the master, and then manually trigger a failover while monitoring KPIs for our REST API.

Key Takeaways

The question tests understanding of resilience testing versus security monitoring or database replication, with the common trap being to choose the database failover option when read replicas in Cloud SQL do not provide failover capability.

This question tests how to properly perform resilience testing on an authentication layer composed of a regional managed instance group and a Cloud SQL database. The correct approach is to simulate a zone-level failure by shutting down all VMs in one zone, which directly validates whether the regional MIG can withstand and recover from compute resource loss.

Common mistake: Many learners choose option D, believing that manually triggering a Cloud SQL read replica failover constitutes resilience testing. However, Google Cloud documentation explicitly states that read replicas do not provide failover capability—they are designed for read scaling, not automatic or manual HA failover to replace a primary instance.

Expert Analysis

Why the Answer Is Correct

Option C is correct because resilience testing specifically evaluates whether a system continues to function despite failures of its components. Shutting off all VMs in a zone simulates a realistic failure scenario and tests whether the regional managed instance group can automatically redistribute traffic to healthy zones. This aligns directly with Google Cloud's guidance on resilience: a resilient application should withstand the loss of one or more compute resources. By monitoring application behavior during this simulated failure, you gain actionable insight into the authentication layer's actual resilience posture.

Why the Other Options Are Wrong

Option A describes credential monitoring services (like Google's Dark Web monitoring), which is a security measure, not resilience testing—it does not test how the system behaves under failure conditions. Option B deploys intrusion detection software, which addresses unauthorized access detection, a security concern rather than a resilience or availability concern. Option D proposes configuring a Cloud SQL read replica and manually triggering failover; however, Google's documentation explicitly states that read replicas do not provide failover capability. Additionally, the REST API performs writes, which a read replica cannot serve, making this approach technically unsound for the described workload.

Community Comment Notes

Multiple commenters converge on C, with vartiklis clarifying the distinction between testing authentication itself versus testing "the resilience of the authentication layer" by simulating component failure. PeppaPig initially argues for D, reasoning that the database is the single point of failure, but Tamim321 correctly counters that "Read replicas do not provide failover capability" per Google's own documentation. juliansierra further notes that because the REST API reads and writes, a read replica alone cannot handle writes, reinforcing that D is technically flawed. Kri_2525 anchors the answer in Google's scalable and resilient apps documentation, where resilience is defined as the ability to withstand unexpected failures.

Official Reference

These links open the official vendor documentation in a new tab.

Exam Strategy

When you see 'resilience testing' or 'disaster simulation' in the question stem, focus on options that directly simulate component failure (e.g., shutting down VMs in a zone) rather than monitoring or detective controls. Also verify vendor-specific limitations—Google Cloud read replicas explicitly do not provide failover capability, which eliminates database-failover distractors.

Frequently Asked Questions

Why can't I use a Cloud SQL read replica to test failover for the authentication layer?

Google Cloud documentation states that read replicas do not provide failover capability; they are for read scaling. For HA failover, you need a regional HA configuration with a standby instance in another zone.

Is shutting down VMs in a zone considered proper resilience testing?

Yes. Simulating a zone failure directly tests whether the regional MIG can withstand the loss of compute resources in one zone, which is a core principle of resilience design.

More PCA FAQ →

Discussion 128 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 30 of 128

  1. Kri_2525 👍 66
    As per google documentation(https://cloud.google.com/solutions/scalable-and-resilient-apps) answer is C. C: A well-designed application should scale seamlessly as demand increases and decreases, and be resilient enough to withstand the loss of one or more compute resources. Resilience: designed to withstand the unexpected A highly-available, or resilient, application is one that continues to function despite expected or unexpected failures of components in the system. If a single instance fails or an entire zone experiences a problem, a resilient application remains fault tolerant—continuing to function and repairing itself automatically if necessary. Because stateful information isn’t stored on any single instance, the loss of an instance—or even an entire zone—should not impact the application’s performance.
  2. vartiklis 👍 20
    You're not testing authentication, you're testing the resilience of the authentication layer. "A resilient app is one that continues to function despite failures of system components" (https://cloud.google.com/architecture/scalable-and-resilient-apps#resilience_designing_to_withstand_failures) - such as shutting down all VMs in a zone.
  3. KouShikyou 👍 16
    Since the question is asking to do a resilience testing, I prefer C.
  4. PeppaPig 👍 9
    D is correct. Regional MIG by default ensures resilience of VMs on application layer What you really need to test is the Database layer as there is only a single DB instance that's obviously became single point of failure. Manual failover means you promote a replica as a standalone database, it is a recommended practice by GCP in disaster recovery. Ref: https://cloud.google.com/sql/docs/mysql/replication/cross-region-replicas#promote-a-replica
  5. SMS 👍 8
    Ignore above. Answer is C
Show 25 more comments
  1. Paul_DSouza 👍 7
    The word "resiliency" refers to =============== To test resiliency, you should verify how the end-to-end workload performs under intermittent failure conditions. =============== Under this:- HA - VMs in different zones to prevent app going down when one zone goes down DR - Entire app infra moves to another region with minimum downtime Testing resiliency assumes a degree of HA or DR or both. C is the most proper answer in this case, as it will test resiliency (of everything including authentication) by simulating a downtime (shutting down all VMs in one zone).
  2. omermahgoub 👍 6
    The correct answer is D. Configure a read replica for your Cloud SQL instance in a different zone than the master, and then manually trigger a failover while monitoring KPIs for our REST API. Resilience testing is a process of evaluating the ability of a system to recover from failures or disruptions. In this case, the customer has an authentication layer consisting of a regional managed instance group serving a public REST API that reads from and writes to a Cloud SQL instance. To perform resilience testing, one option would be to configure a read replica for the Cloud SQL instance in a different zone than the master. This would allow you to test the system's ability to recover from a failure or disruption in one zone. Then, you could manually trigger a failover while monitoring key performance indicators (KPIs) for the REST API, such as response times and error rates. This will allow you to see how the system behaves during a failover and identify any potential issues that need to be addressed.
  3. juliansierra 👍 6
    I choose C. I don't say D because the REST API read and WRITE in the database, if you create a READ replica in Cloud SQL, the REST API will not have the possibility of write in the database. The answer D doesn't mention anything about promote the read replica to master.
  4. Tamim321 👍 6 Answered C
    Read replica do not provide failover capability https://cloud.google.com/sql/docs/mysql/replication#:~:text=Note%3A%20Read%20replicas%20do%20not,HA%20on%20your%20primary%20instance.
  5. Eroc 👍 6
    I'm with @jcmoranp and @crypt0 on this one... shutting down the VMs wouldn't test only the authentication layer... same for replication of database (D).. answer B
  6. Jack_in_Large 👍 6
    Shutting off all VMs in a zone is not good approach for testing of authentication
  7. JC0926 👍 5 Answered D
    D. Configure a read replica for your Cloud SQL instance in a different zone than the master, and then manually trigger a failover while monitoring KPIs for our REST API. This option is the most suitable for resilience testing of the authentication layer. By configuring a read replica in a different zone than the master, you add redundancy to your system. Manually triggering a failover while monitoring KPIs for your REST API helps you observe how the system behaves during a failure and ensures that the authentication layer remains available and performs as expected during an outage or disaster.
  8. nrajesh 👍 5
    C is the answer. D is misleading. Read replica does not provide failover capability. If it was failover replica, we would have a tie between C and D https://cloud.google.com/sql/docs/mysql/replication/create-replica -"Note: Read replicas do not provide failover capability. To provide failover capability for an instance, see Configuring an Instance for High Availability."
  9. PeppaPig 👍 5
    I would go with D. Obviously a single DB instance would become a single point of failure. You can prompt a cross-region read replica to the primary in the case of DR. https://cloud.google.com/sql/docs/mysql/replication/cross-region-replicas
  10. nezih 👍 5
    this is about testing resilience not improving. So deploying IDS won't give any idea how resilient our system is. Answer is C
  11. Unfaithful 👍 5
    Answer: C Support: Read replicas do not provide failover capability. So option D is out.
  12. gfhbox0083 👍 5
    C, for sure. When talking about Resilience.
  13. KouShikyou 👍 5
    How about D?
  14. MaxNRG 👍 4
    C – schedule a disaster simulation exercise during which you can shut off all VMs in a zone to see how your app behaves. A, B – don’t test resilience at all, just security / authentication attacks. But, Q asks different – check if authentication works if some resources are down. So, C and D to consider. Resiliency definition: a highly-available, or resilient app is one that continues to function despite expected or unexpected failures of components in the system. D – says about Read Replica and trigger a failover. But, Read replicas neither provide High-Availability nor offer it; a master instance cannot failover to a read replica, and read replicas are unable to fail over in any way during outage. C – focuses on both aspects: resilience and authentication layer testing by shutting down FrontEnd VMs in the zone. Cloud SQL testing is OOO, since this is back-end / data layer.
  15. omermahgoub 👍 4
    Option A, engaging with a security company to run web scrapers looking for users' authentication data on malicious websites, is not related to resilience testing. Option B, deploying intrusion detection software to your virtual machines to detect and log unauthorized access, is a good practice for improving the security of your system, but it is not directly related to resilience testing. Option C, scheduling a disaster simulation exercise during which you can shut off all VMs in a zone to see how your application behaves, is a valid method of resilience testing, but it does not involve testing the system's ability to recover from a failure or disruption in one zone.
  16. ybe_gcp_cert 👍 4
    A and B focus on security, intrusion test and scraper scan doesn't test the resiliency which is a broader domain. Which leave C and D. D might do the job but, the problem is changing the configuration by adding a replica. You are supposed to design your application and test it's resilience as is, not add new components. The question doesn't ask about what can you do to improve or implement a DR. it's asking to DO resilience test. I would go with C.
  17. MeasService 👍 4
    So... again very confusing question. They highlighted first of Authentication layer resilience testing. but.... they continue saying..... THIS CONSIST of regional MIGs and Cloud SQL. When I hear on Resilience testing..... I would look for Is service available when some components down. That leaves me with option C and D. and out of both... C would be one I would go with.
  18. Sat_G 👍 4
    D - the question says resilience testing of authentication layer. Its using REST API with Cloud SQL ... Cloud SQL could very easily be the layer that does authentication hence having a read replicate and manually trigger a fail over proves the objective. C is an overkill to turn off all your VMs to test authentication resiliency.
  19. GunjGupta 👍 4
    There are 2 key points in Question-- Resiliency testing and authentication layer. The way I read the question, the Authentication layer consists of MIG with Cloud SQL. Hence we need to propose a solution for creating chaos on this (MIG with Cloud SQL). A and B do not seems to reflect resiliency aspects. I will go with C
  20. kalschi 👍 4
    I think the question states the _authentication_layer_ consists of regional MIG and Cloud SQL. so to _resilience_test_ the _authentication_layer_ , it's really to test the MIG and Cloud SQL. So prefer C. Also, intrusion detection software is not TESTING the layer, it's more like a detection and prevention process.
  21. ronald89 👍 4
    D should not be the answer, since "This consists of a regional managed instance group serving a public REST API that reads from and writes to a Cloud SQL instance" ; D is creating a read-only instance.
  22. crypt0 👍 4
    I am also not sure here, but they say: resilience testing of their _authentication layer_. How does shutting down all vms from one zone test the authentication layer?
  23. fiercedog 👍 4 Answered D
    Answer is D. App layer is covered by regional managed instance group. Database is the point of failure that needs to be addressed.
  24. victory108 👍 4
    C. Schedule a disaster simulation exercise during which you can shut off all VMs in a zone to see how your application behaves.
  25. MeasService 👍 4
    Is it really B? why not C ? It is taking about resilience testing. shutting down all vms from one zone would be ideal testing.

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Multi-Regional vs Regional Cloud Storage for Global Software Downloads

S-grade

Surviving a Zone Outage With SQL Server on GCP

S-grade

Fixing BigQuery connection errors from a Compute Engine VM

S-grade

Moving 75 TB into Cloud Storage with Google-recommended practices

S-grade

Practice All PCA Questions

Access 81 questions with complete answers and detailed explanations.

View Full PCA Practice Test →

← Back to PCA Study Guide