The requirement is reliance testing (recovery from failure )while maintaining 99.9 availability (SLA)

Let's go with elimination.

A. Capture existing users input, and replay captured user load until autoscale is triggered on all layers. At the same time, terminate all resources in one of the zones

>> We are overkilling resilience testing by entering the boundary of DR hence will eliminate this option.

B. Create synthetic random user input, replay synthetic load until autoscale logic is triggered on at least one layer, and introduce ג€chaosג€ to the system by terminating random resources on both zones

>> This option takes care of the requirement of testing. Is a patter used by org like Netflix chaos monkey).

C. Expose the new system to a larger group of users, and increase group size each day until autoscale logic is triggered on all layers. At the same time, terminate random resources on both zones

>> Next quarter your user going to increase, but how are you going to get more users to test, this option is not feasible hence eliminating.

D. Capture existing users input, and replay captured user load until resource utilization crosses 80%. Also, derive estimated number of users based on existing userג€™s usage of the app, and deploy enough resources to handle 200% of expected load

>> The failure is not necessary from CPU utilization, even the hardware can fail under load. Blanket provisioning is not the way to go hence eliminating.

Hence B is correct answer.

C & E:

C: Smaller the base image with minimum dependency faster the container will start

E: Docker image build uses caching. Docker Instructions sequence matter because

application’s dependencies change less frequently than the Python code which will help to reuse the cached layer of dependency and only add new layer for code change for Python Source code.

Question talks about performance issue, D would not help in identifying performace issue because it would make available to all users , C would be correct answer ,as we have to indentify performace issue.

D is correct, the headline in GC trace documentation says it all: "Cloud Trace is a distributed tracing system for Google Cloud that collects latency data from applications and displays it in near real-time in the Google Cloud Console."

Answer is D. The problem statement is that replica server not got promoted as master. There is nowhere mentioned the reason for primary DB going down because of resource contention (performance issue). So focussing on the problem statement D is the correct answer

Answer D is correct. Again look for the keywords. Analysis only when Possible legal proceedings. It's cheaper to keep the data in Cloud Storage for long term rather than keeping it on the BQ. Also, when there is a possible legal proceeding happens the Cloud Storage data can be fed to BQ and do the analysis. It will be cheaper. So most cost effective is 'D'. The analysis of the data requirement is confusing if not read properly.

Let's go with option elimination

A. Google Cloud Dedicated Interconnect

>> Secured, fast connection, hence the choice. This will allow private connection from GCP to the data centre with a fast connection. Cost is not mentioned in the requirement to eliminate this option.

B. Google Cloud VPN connected to the data centre network

>> We have to think about data flowing on the internet and the requirement talks about private connect. Also not sure how well you connect VPN with Data Center until you use the hybrid option. https://cloud.google.com/network-connectivity/docs/vpn/concepts/overview hence eliminate

C. A NAT and TLS translation gateway installed on-premises

>>This is a VM option to reach outside won't for this requirement hence eliminate

D. A Google Compute Engine instance with a VPN server installed connected to the data centre network

>>This is a slow option hence eliminate

Hence A

B is correct answer.

Export to Bigquery and restrict access to queries with ACLs to auditors.

C is the answer, since key management systems generate, use, rotate, encrypt, and destroy cryptographic keys and manage permissions to those keys.

A is incorrect because storing credentials in source code and source control is discoverable, in plain text, by anyone with access to the source code. This also introduces the requirement to update code and do a deployment each time the credentials are rotated.

B is not correct because consistently populating environment variables would require the credentials to be available, in plain text, when the session is started.

D is incorrect because instead of managing access to the config file and updating manually as keys are rotated, it would be better to leverage a key management system. Additionally, there is increased risk if the config file contains the credentials in plain text.

A. Nothing to see

B. Any functionality of any tool can be deprecated (Terraform, Chef ...)

C. It's true

D. Any tool needs some kind of credential to run

E. It's true