Cloud MLOps Intro Quiz

How can Docker be integrated into a microservices architecture for CI/CD pipelines?

• • By using Docker containers as lightweight, standalone services.

• • For version control of microservices.

• • As a replacement for virtual machines.

• • Docker is not suitable for microservices architectures.

What is the best practice for handling sensitive data in Docker containers?

• • Storing them as environment variables in Dockerfile.

• • Using Docker secrets or external secrets management tools.

• • Embedding them in the application code.

• • Passing them as command-line arguments during container runtime.

In Docker, how do layers work in an image and what is their impact on image building and sharing?

• • Layers are irrelevant in Docker images.

• • Each layer is a modification of the image, which reduces the image size.

• • Layers represent changes in the image, which improves sharing and reduces build time by reusing layers.

• • Layers increase the security of Docker images.

What are the key considerations when designing a scalable serverless architecture using AWS Lambda in terms of scalability and cost?

• • Unlimited scalability and fixed cost.

• • Scalability is manually managed, with variable cost.

• • Automatic scalability, with cost scaling linearly with usage.

• • No scalability, but cost-effective for large applications.

How would you optimize data transfer to S3 for high-latency networks?

• • Using Amazon Glacier instead of S3.

• • Enabling S3 Transfer Acceleration.

• • Increasing the size of data packets.

• • Switching to a different AWS service like EBS.

What is the significance of edge locations in AWS CloudFront, and how do they enhance content delivery?

• • They are central data storage locations for AWS services.

• • Edge locations reduce the cost of content delivery.

• • They cache content closer to users, reducing latency.

• • Edge locations are used for AWS administrative tasks.

What is the purpose of the @staticmethod decorator in Python?

• • To create a method that requires an instance of its class to be called.

• • To define a method that can modify the class state without requiring an instance.

• • To declare a method that does not require an instance or a class reference to be called.

• • To ensure that a method is thread-safe and can be accessed by multiple threads simultaneously.

When dealing with a large dataset joined with a tiny dataset in Apache Spark, what is the best practice to optimize the join operation and why?

• • Use a regular join as Spark automatically optimizes all join operations.

• • Partition the larger dataset to match the number of partitions in the smaller dataset.

• • Broadcast the smaller dataset using Spark's broadcast variable feature.

• • Convert the smaller dataset to an RDD and use an RDD join operation.

In a Spark application, how does the small file problem impact processing efficiency and what are common strategies to mitigate this issue?

• • The small file problem increases parallelism, which is mitigated by consolidating files before processing.

• • It causes excessive metadata overhead, which can be mitigated by merging small files into larger ones before processing.

• • Small files increase the speed of data processing; no mitigation is needed.

• • The problem reduces data redundancy, solved by creating multiple copies of the small files.

In Continuous Deployment, what is a common strategy for handling a failed deployment?

• • Pausing all further deployments until the issue is manually resolved.

• • Implementing an automated rollback to the last stable version.

• • Increasing the number of server instances to handle the failure.

• • Permanently removing the feature that caused the failure.