Exam Cloud Digital Leader topic 1 question 47 discussion

A for me, with CUD you get a greater discount and it makes totally sense to buy a CUD for the baseline traffic since it will always be consumed

Will gop with C. as CUD is applicable for 1 or 3 year period. In the ques, there is no mention of that info. So SUD is appropriate as we cannot assume that the VMs will be used for 1 or 3 years..

Commited use discount for base is cost effective and it doesnt restrict spiked usage

Commited use discount for base is cost effective and it doesnt restrict spiked usage

Correct answer is: A. Purchase committed use discounts for the baseline load. Explanation: Committed Use Discounts (CUDs) are ideal for predictable workloads. By purchasing CUDs for the baseline load, your organization can secure significant savings on the resources that will be consistently used, which helps manage costs effectively. This approach allows you to commit to a certain level of usage over a specified period (usually one or three years) and receive discounted rates compared to on-demand pricing. Running the workload on preemptible VM instances (option D) could save costs, but preemptible VMs can be terminated at any time, which may not be suitable for critical workloads that require reliability.

A for me .

for regular spike, its better to go for committed baseline

D. Run the workload on preemptible VM instances Here’s why this is the best option: Preemptible VM instances: These are short-lived, low-cost instances that are ideal for handling temporary, bursty workloads. They are significantly cheaper than standard VM instances but can be terminated by Google Cloud with little notice (about 30 seconds). For workloads with predictable spikes, using preemptible VMs can provide a cost-effective solution to handle the increased demand without incurring high costs.

The best option to control cost for your organization's image processing application is: A. Purchase committed use discounts for the baseline load Here's why the other options are less suitable: B. Purchase committed use discounts for the expected spike load: Committed use discounts are most beneficial for predictable, ongoing resource usage. Since the spike load is temporary, committing to those resources wouldn't be cost-effective. C. Leverage sustained use discounts for your VMs: Sustained use discounts reward consistent resource usage at a discount compared to on-demand pricing. However, they might not be ideal for workloads with significant spikes that could push you outside the sustained use window. D. Run the workload on preemptible VM instances: Preemptible VMs are a good option for highly flexible workloads that can be interrupted. However, for image processing tasks that take hours, interruptions during processing could lead to wasted work and require restarts.

your workload has a predictable baseline level. Google Cloud offers sustained use discounts for Compute Engine commitments that reserve VM instances for a period of 1 year or 3 years. This can significantly reduce your costs for the baseline usage.

Sustained use discounts provide cost savings for running virtual machines continuously over a month. Since the workload normally stays at a certain baseline level, leveraging sustained use discounts can help reduce costs for the consistent usage. This approach is suitable when the workload is predictable and doesn't involve significant spikes. However, if the expected spike load is known in advance, option B: Purchasing committed use discounts for the expected spike load could also be considered to optimize costs during those periods. It is important to analyze the workload patterns and cost implications to make an informed decision.

C for me

C is correct

A would be correct, since "controlling the cost" is a requirement

I think it should be A.

C is correct

why not to purchase committed use discounts for the baseline load. Here are the reasons why this is not a good option: It would be more expensive than sustained use discounts because you would be paying for the baseline load even when you are not using it. It would not take into account the spike load. If the spike load is significant, then you would still be paying a lot for the resources that you are using during the spike load. It would be inflexible. If your workload changes, then you would have to change your committed use commitments, which can be difficult and time-consuming.