1Z0-1067-25 Question 46
Select 2Your company runs a CPU-intensive batch job in OCI that processes large data sets for approximately six hours every night. Outside of this window, the compute instance remains idle but continues to incur costs. You want to reduce overall expenses without sacrificing the job's performance. Which two strategies could help you optimize cost and still meet your performance requirements?
- A
Use instance scheduling to start and stop the compute instance according to the batch window.
- B
Switch to a flexible shape (e.g., VM.Standard.E3.Flex) and allocate the exact number of OCPUs and memory required for the job.
- C
Store the batch output on ephemeral local NVMe volumes to reduce storage costs and improve IOPS.
- D
Deploy the batch job on a dedicated host to avoid sharing compute resources with other tenants.
- E
Migrate the batch process to Oracle Container Engine for Kubernetes (OKE) and use a fully-managed cluster running 24/7.
Show answer and explanation
Correct answers: A, B
Explanation
By scheduling instance start/stop cycles, you only pay for compute during the actual usage period, and using flexible shapes helps you further optimize the number of OCPUs and memory according to your batch job� requirements. For data durability, persistent storage remains necessary. For additional details, consult Oracle� documentation on 'Instance Scheduling' and 'Flexible Compute Shapes' to orchestrate resources effectively while minimizing idle spending.
- A. Correct.
Correct. Using instance scheduling (start/stop) ensures you only pay for the compute resources during the six-hour batch window. Outside of that time, the instance is stopped, eliminating unnecessary costs.
- B. Correct.
Correct. Flexible shapes allow you to tailor the OCPU and memory allocation precisely to your workload, avoiding overprovisioning and reducing cost while maintaining the performance required for your batch job.
- C. Incorrect.
Incorrect. While local NVMe volumes provide high IOPS, they are ephemeral and can be lost if the instance is stopped or fails. This is risky for storing critical batch output. Persistent storage (e.g., Block Volumes or Object Storage) is recommended for durable data.
- D. Incorrect.
Incorrect. Dedicated hosts are typically used to meet specific compliance or licensing requirements and can be more expensive. This does not inherently reduce costs for a short-term batch workload that only runs a few hours a day.
- E. Incorrect.
Incorrect. While containerization can help in other scenarios, running a managed Kubernetes cluster 24/7 without effective node scaling could lead to continuous charges. If you cannot effectively scale the worker nodes down to zero, this approach may not provide net cost savings compared to stopping VM instances.