Відео

Were the questions on your exams?

suppose we upload a file in S3 bucket and we want the same file to be received over mail. Then what to do. Please reply 🙏

222 should be C IMO

Thanks, quite helpful

Has anyone given the exam recently in aug or something by studying the questions from this video

159 should be C

Muchas Thanks

video is good and easy to understand. Thankyou! ☺

Question 306 should be reviewed

Question 274 - The best answer is C. Log debug information in each ParDo function, and analyze the logs at execution time. Here's why: The problem: The pipeline is taking too long to process data, causing output delays. This suggests a bottleneck somewhere within the pipeline's processing steps. Why other options are less suitable: A. Insert a Reshuffle operation...: While reshuffling can sometimes help with performance, it's not a direct way to identify the bottleneck. It's more of a general optimization technique. B. Insert output sinks...: This might help monitor the overall throughput, but it doesn't pinpoint the specific step causing the delay. D. Verify that the Dataflow service accounts...: This addresses potential permission issues, but it's unlikely to be the root cause of a processing delay. Why C is the best solution: Directly addresses the problem: Logging debug information within the ParDo functions (which are the individual processing units in Dataflow) provides granular insights into the execution of each step. Identifies the bottleneck: By analyzing the logs, you can see which ParDo function is taking the longest to execute, revealing the bottleneck.

Question 272 - The best answer is C. Use BigQuery Data Transfer Service with the Teradata Parallel Transporter (TPT) tbuild utility. Here's why: Efficiency: The TPT tbuild utility is specifically designed for efficient data transfer between Teradata and other systems. It leverages parallel processing to speed up the export process. Minimal Programming: The tbuild utility is a command-line tool that requires minimal coding, making it ideal for reducing development effort. Storage Constraints: The TPT tbuild utility allows you to specify the output format and location, enabling you to directly transfer data to Cloud Storage without the need for intermediate local storage, addressing the limited storage space constraint. Step-by-Step Solution: Configure TPT: Install and configure the Teradata Parallel Transporter (TPT) on your Teradata server. Create TPT Script: Use the TPT tbuild utility to create a script that exports the historical data from Teradata. Specify Output Format: In the TPT script, specify the output format as a supported BigQuery format (e.g., Avro, Parquet). Direct Transfer to Cloud Storage: Configure the TPT script to directly transfer the exported data to Cloud Storage. BigQuery Data Transfer Service: Set up a BigQuery Data Transfer Service instance to load the data from Cloud Storage into your BigQuery tables. Graphical Representation: Teradata Server | | TPT tbuild utility | (Exports data to Cloud Storage) | | Cloud Storage | | BigQuery Data Transfer Service | | BigQuery Tables Advantages of this approach: Parallel Processing: TPT tbuild utilizes parallel processing for faster data export. Direct Transfer: Data is directly transferred to Cloud Storage, eliminating the need for local storage. Simplified Workflow: The process is streamlined with minimal coding and a clear path for data movement. Built-in Integration: BigQuery Data Transfer Service seamlessly integrates with Cloud Storage for efficient loading.

Question 267 - The correct answer is **D. 1. Enable the Airflow REST API, and set up Cloud Storage notifications to trigger a Cloud Function instance. 2. Write a Cloud Function instance to call the DAG by using the Airflow REST API and the web server URL. 3. Use VPC Serverless Access to reach the web server URL. ** Here's why: Cloud Composer and Subnetwork: Your Cloud Composer instance is in a subnetwork with no internet access. This means you cannot directly access the web server URL from the cloud storage notifications. Cloud Function: A Cloud Function is a serverless compute environment that can be triggered by events like Cloud Storage notifications. This allows for a reactive approach to DAG execution. Airflow REST API: The Airflow REST API provides a way to programmatically trigger DAGs. This is essential for our Cloud Function to execute the DAG. VPC Serverless Access: This feature allows serverless functions like Cloud Functions to securely access resources in a private network like your Cloud Composer instance. Here's a breakdown of the steps and a diagram to illustrate: 1. Enable Airflow REST API This exposes an API endpoint that allows you to manage and trigger DAGs. 2. Set up Cloud Storage Notifications Configure Cloud Storage to trigger a Cloud Function whenever a new file is added to the bucket. 3. Create a Cloud Function This function is triggered by the Cloud Storage notification. Within the function: It makes an API call to the Airflow REST API using the web server URL. The function specifies the DAG name to be executed. It uses VPC Serverless Access to securely connect to the private network where Cloud Composer is running. 4. VPC Serverless Access This allows your Cloud Function to connect to your private network securely. Diagram: ┌──────────────────────┐ │ Cloud Storage Bucket │ └──────────────────────┘ ^ | | Cloud Storage Notification | ↓ ┌──────────────────────┐ │ Cloud Function │ └──────────────────────┘ ^ | | API Call | ↓ ┌──────────────────────┐ │ Airflow REST API │ └──────────────────────┘ ^ | | VPC Serverless Access | ↓ ┌──────────────────────┐ │ Cloud Composer Instance│ └──────────────────────┘ └── DAG Execution Why other options are not suitable: Option A: Enabling Private Google Access in the subnetwork and setting up Cloud Storage notifications to a Pub/Sub topic wouldn't work because the Cloud Composer instance has no internet access, and Pub/Sub topics are not designed to directly trigger DAG executions. Option B: While enabling the Cloud Composer API could potentially allow you to trigger DAGs, using a Cloud Function and VPC Serverless Access is a much more secure and efficient way to handle communication between your serverless environment and a private network. Option C: Using a Private Service Connect (PSC) endpoint would be a more complex approach. The Cloud Function would need to establish a connection to the PSC endpoint, making the communication more involved. In summary, option D is the most straightforward and efficient way to achieve reactive DAG execution in your Cloud Composer instance with no internet access.

Question 264 - The best answer is C. Create a VPC Service Controls perimeter containing the VPC network and add Dataflow, Cloud Storage, and BigQuery as allowed services. Use Dataflow with only internal IP addresses. Here's why: VPC Service Controls provide a robust way to enforce security policies within your VPC network. Perimeter allows you to define a set of resources and services that are allowed to communicate with each other, effectively creating a secure boundary. Internal IP Addresses are the key to achieving the organizational constraint of using only internal IP addresses for Compute Engine instances. Let's examine why the other options are less suitable: A. Ensure your workers have network tags to access Cloud Storage and BigQuery. Use Dataflow with only internal IP addresses. While network tags can help with access control, they don't guarantee secure communication within your VPC. B. Ensure that the firewall rules allow access to Cloud Storage and BigQuery. Use Dataflow with only internal IPs. Firewall rules alone aren't enough to enforce the organizational constraint, as other settings can override them. D. Ensure that Private Google Access is enabled in the subnetwork. Use Dataflow with only internal IP addresses. While Private Google Access helps access Google services privately, it doesn't encompass the entire security policy enforced by the organization. Therefore, Option C is the most comprehensive and effective solution to address the organizational constraint and maintain security within your Dataflow pipeline.

Question 239 - B and E Here's why: B. Use Pub/Sub Snapshot capture two days before the deployment. This is the most direct way to ensure reprocessing. You create a snapshot of the subscription data at a point in time (two days before deployment). After the update, you can use this snapshot to replay messages for the past two days, allowing your updated pipeline to handle them. E. Use Pub/Sub Seek with a timestamp. You can use Seek with a timestamp of two days in the past to tell your pipeline to start reading messages from that point forward. This way, all the messages from the previous two days will be processed by your updated pipeline.

I have doubt.. Can we able to directly move to the s3 without RDS in incremental method

Why you are using RDS? We can directly move the SQL server data to the S3

very useful video. could you please also upload tutorial for etl testing having more than 150 procedures and 400+ reports (from sybase to mssql)

Question 170 - The best answer is C. Cloud Spanner. Here's why: Cloud Spanner is a fully managed, globally distributed, relational database service that offers: Automatic scaling Transactional consistency Ability to handle large datasets (including up to 6TB) Querying with SQL Let's look at the other options: Cloud SQL: While it's fully managed and offers SQL querying, it doesn't provide the same level of global distribution and automatic scaling as Cloud Spanner. Cloud Bigtable: This is a NoSQL database, not a relational database, and it wouldn't be the best fit for SQL-based querying. Cloud Datastore: Similar to Bigtable, it's a NoSQL database. Cloud Spanner best fits the requirements of this scenario, offering a fully managed, scalable, and globally consistent relational database.

Hey this was really helpful I cleared my exam thanks can you please upload questions for gcp professional cloud architect

Question 131 - C Here's why: Creating a new pipeline with the same Pub/Sub subscription ensures that data continues to flow into the new pipeline, preventing data loss. Canceling the old pipeline once the new one is running and stable ensures you're not running two pipelines unnecessarily and potentially leading to confusion. Let's analyze why the other options are incorrect: A. Update the current pipeline and use the drain flag: While this might work, it could lead to complications if the code changes make the old pipeline incompatible with the new version. This approach might not always guarantee data loss prevention. B. Update the current pipeline and provide the transform mapping JSON object: This approach doesn't address the core issue of the code change making the pipeline incompatible. It focuses on transformation mapping, which doesn't inherently prevent data loss if the pipeline structure changes significantly. D. Create a new pipeline that has a new Cloud Pub/Sub subscription and cancel the old pipeline: This would cause data loss as the new pipeline would not receive messages from the old subscription. In summary, the safest and most efficient solution is to create a new pipeline with the same Pub/Sub subscription to maintain data flow and then cancel the old pipeline once the new one is fully operational.

Question 89 - The C option instead of D RMSE (Root Mean Squared Error) is a common metric used in linear regression to measure the difference between the predicted values and the actual values. It essentially represents the average magnitude of the errors. A lower RMSE indicates a better fit of the model to the data, meaning the model is making more accurate predictions. In this question, the model has higher RMSE on the test set than the training set, which indicates the model is overfitting. This means it is performing well on the training data but struggles to generalize to new data. The best answer to improve the performance is C. Try out regularization techniques (e.g., dropout of batch normalization) to avoid overfitting. Regularization techniques help prevent overfitting by adding constraints or penalties to the model's learning process. These techniques effectively "shrink" the model's parameters and discourage it from fitting the training data too closely, leading to better generalization on unseen data.

Question 88 - The D option instead of C The best answer is C. Use Cloud GPUs after implementing GPU kernel support for your customs ops. Here's why: TPUs (Tensor Processing Units) are optimized for large matrix multiplications, which are common in deep learning. However, TPUs are generally more expensive than GPUs. GPUs (Graphics Processing Units) are also very capable for matrix multiplications and are generally more cost-effective than TPUs. Implementing GPU kernel support for custom C++ TensorFlow ops is crucial. This allows your custom ops to leverage the parallel processing power of GPUs, significantly accelerating training. Option D is not the best solution because increasing the size of the cluster only increases the cost without necessarily providing a significant speedup. While adding more CPUs could help, it won't be as efficient as utilizing specialized hardware like GPUs. In summary, using GPUs with optimized custom ops is the most efficient way to achieve faster training times and lower costs when dealing with matrix-heavy workloads.

Question 54 -The B option instead of C, no? Here's why - my explanation : Multi-cluster routing ensures that your production application continues to receive fast responses, even with the added analytical workload. The production application will primarily use the original cluster, and the analytical workload will primarily use the new cluster. Live-traffic app profile for the production application maintains optimal performance for real-time operations. Batch-analytics profile for the analytical workload allows for the efficient processing of large data sets.

Regarding Question #3 @3:22: IMO - The correct answer should be B - Enable auto-discovery of files for the curated zone. Reason: In Dataplex assets should be setup with the correct file formats and paths to enable automatic discovery. So one has to verify the assets configuration settigs to ensure they align with the given file types (JSON and CSV). Also Dataplex uses auto-discovery to detect and catalog new files, so we have to configure auto-discovery settings or triggers to ensure that new files are picked up automatically. Please suggest or provide more details why option A (moving files to raw zone) is right answer. This will help me and others to understand it more better.

Question 48 - My answer is B - Switch to TFRecords formats (appr. 200MB per file) instead of parquet files. Explanation TFRecords are a binary file format designed specifically for storing and efficiently reading data for TensorFlow models. They are typically smaller and more optimized for loading and processing, which can lead to performance improvements compared to Parquet files, especially in distributed systems like Spark jobs. Option A is incorrect because increasing the size of Parquet files would likely lead to more network overhead and potentially slower processing times. Option C is incorrect because, although switching to SSDs can improve performance, copying data to HDFS adds an unnecessary step and could introduce additional latency, and using a preemptible VM config may actually reduce performance in the long run. Option D is incorrect because overriding the preemptible VMs configuration could make the system less cost-effective and could potentially lead to data loss if a preemptible VM is terminated prematurely. Key Takeaways: Using TFRecords can optimize your Spark jobs on Dataproc, especially for large-scale analytical workloads. Preemptible VMs are cost-effective, but their usage should be carefully considered in terms of potential performance impacts. Remember that for large datasets, using more preemptible VMs with optimized TFRecord formats is more effective than increasing the boot disk size of the VMs and trying to accommodate larger Parquet files.

I think question 46.. The best answer is D. Increase the amount of concurrent slots per project at the Quotas page at the Cloud Console. Here's why: Direct Solution: This option directly addresses the problem of users not getting slots to execute their queries. Increasing the concurrent slot quota for each project allows more users to run queries simultaneously, reducing the risk of resource limitations. Let's examine why the other options are not ideal: A. Convert your batch BQ queries into interactive BQ queries: While this might help with some scenarios, it's not a guaranteed solution. Interactive queries can be more resource-intensive, and it might not address the root issue of insufficient slots. B. Create an additional project to overcome the 2K on-demand per-project quota: Creating new projects can be complex and might lead to organizational inefficiencies. It doesn't solve the underlying issue of limited slots, and could potentially lead to more projects competing for resources. C. Switch to flat-rate pricing and establish a hierarchical priority model for your projects: Flat-rate pricing can be expensive and might not be necessary for all projects. Establishing a hierarchical priority model could introduce its own complexities and might not solve the immediate problem of users not getting slots. Key Takeaway: In this scenario, the most efficient and effective solution is to directly increase the available concurrent slots for each project. This allows more users to run queries without introducing additional complexity or cost.

About question 37 I think option D is the correct answer - The best answer is D. In a bucket on Cloud Storage that is accessible only by an AppEngine service that collects user information and logs the access before providing a link to the bucket. Here's why: Auditable Record: AppEngine's logging of access attempts directly provides an auditable record. The service collects user information and logs access attempts, creating a detailed history. Security: The fact that the bucket is accessible only through AppEngine adds a layer of security. Only authorized AppEngine instances can access the data, preventing unauthorized access. Link to Bucket: Providing a link to the bucket ensures that the AppEngine service controls access, making it possible to restrict access to the data based on specific criteria or authorizations. The other options have drawbacks: A: Encrypted on Cloud Storage: While encryption is good, user-supplied keys can be less secure and don't automatically provide an auditable access log. B: In a BigQuery dataset: BigQuery's Data Access log can provide some auditability, but it doesn't specifically track individual user access. C: In Cloud SQL: Cloud SQL's Admin activity logs track administrative changes, not user-level access to the data. In summary, option D offers the most secure and auditable way to store data while meeting the mandate of an auditable access record.

Q311 answer is B or D?

Thank you so much! It worked, one table's DDL got updated and later I wanted to recover it.

Hi, can you make a video of latest quetions.

question 109, answer D should have ` as the first character instead of '

Is question 29 up to date? I'm reading Stackdriver was discontinued in Feb 2020

Hello, do you have the questions for the machine learning certification?

Hi sir, Database migration task is running , but updated data is not reflected in s3 what may be the reason

thank you so much

Cleared my exam today, because of this playlist, thanks

Paased my exam today, thanks for your efforts :)

please make video on Google Cloud ACE certification also

question 227 - I think the correct answer is option A

question 15 I think option D is the correct answer

how can we do the similar action in Glue using Visual ETL approach

You don't have the pdf with questions? Thank you so much!

question 3 : option B no ?

hey guy thanks i was getting mad because didn't know how to pass a parameter

can you plz provide ms-replication commands which is required on ec2 to rds migration process

Thanks for the videos ..i passed the exam today ...great effort man...much appreciated 👍

are these latest questions patterns, or these are like advanced level prepartion for the exam so that actual exam would be easy ?

Thank you, I just passed the exam, many of your questions came to the exam

Test Endpoint failed: Application-Status: 1020912, Application-Message: Failed to connect Network error has occurred, Application-Detailed-Message: RetCode: SQL_ERROR SqlState: 08001 NativeError: 101 Message: [unixODBC]FATAL: password authentication failed for user "user". When I am trying to do connection test for source Endpoint, I am getting the above error. But I am able to connect to the db from DMS query Editor. What am i doing wrong.

Can you please share the pdf version of the question