Domain 5: Implementation & Reliability

**CASE STUDY: TechStream Gaming** **Company Overview:** TechStream Gaming is a global gaming company with 500 employees and $50M annual revenue. **Current Environment:** On-premises data centers in US and EU. 200 servers. MySQL databases (5 TB). Peak users: 2M. Cost: $100K/mo. **Business Requirements:** Reduce costs by 40%. Support 5x user growth. Launch in APAC, SA, Africa. Improve deployment to daily. **Executive Statements:** CEO: 'Scale rapidly.' CFO: 'Max $100K/mo, ROI 18mo.' CTO: 'Limited cloud exp, 99.95% uptime.' **Technical Requirements:** <100ms latency globally. Real-time analytics. 5x traffic spikes. EU data residency. DDoS protection. CI/CD. **Constraints:** 12-month migration. Max 4-hour downtime. 20 devs (Java/MySQL), 5 ops (limited cloud). Budget $2M. **QUESTION:** To address the CFO's cost concerns and the CTO's team constraints, which combination of strategies should you implement?

**CASE STUDY: Dress4Win** **Company Overview:** Dress4Win is a web-based retail company that helps users organize their wardrobes. **Current Environment:** Colocated data center. Tomcat app servers, Nginx web servers, MySQL databases, Redis caching. 100TB of image data on SAN. **Business Requirements:** Migrate to cloud to handle seasonal spikes (Black Friday). Reduce CapEx. Enable rapid prototyping. **Executive Statements:** CEO: 'Innovate faster, stop worrying about servers.' CFO: 'Move to OpEx. Ensure PCI-DSS compliance.' CTO: 'Modernize stack but migrate quickly first.' **Technical Requirements:** Secure hybrid connectivity during migration. PCI-DSS compliance. Automated scaling. CI/CD for microservices. **Constraints:** Migration must be completed before Q4 holiday season (6 months). Limited budget for refactoring during initial migration. **QUESTION:** To handle the Black Friday seasonal spikes for the initially migrated Compute Engine workloads, which TWO actions should you take? (Select TWO)

**CASE STUDY: TerramEarth** **Company Overview:** TerramEarth manufactures heavy equipment. 2 million vehicles in the field. **Current Environment:** Vehicles send telemetry via cellular. Processing 100,000 msgs/sec. On-prem Hadoop cluster. **Business Requirements:** Predict equipment failure. Reduce warranty costs. Provide fleet dashboard. **Executive Statements:** CEO: 'Monetize data.' CFO: 'Storage costs spiraling.' CTO: 'Need scalable ingestion and ML.' **Technical Requirements:** Ingest 500,000 msgs/sec. Store petabytes cost-effectively. Train ML models. Real-time anomaly detection. **Constraints:** Intermittent connectivity. Strict vehicle authentication. **QUESTION:** To ensure the reliability of the new ingestion pipeline, the operations team wants to implement SRE practices. How should they measure the performance of the ingestion API?

**CASE STUDY: HealthCare360** **Company Overview:** HealthCare360 provides EHR systems to hospitals in NA and EU. **Current Environment:** Isolated on-prem deployments. Fragmented data. **Business Requirements:** Centralize EHR in cloud. Enable cross-hospital research. Ensure compliance. **Executive Statements:** CEO: 'Transforming to SaaS.' CFO: 'Need cost attribution per tenant.' CSO: 'Zero compromise on HIPAA/GDPR.' **Technical Requirements:** Multi-region active-active deployment. Microservices on GKE. End-to-end encryption (CMEK). Strict network perimeters. **Constraints:** Zero data loss (RPO=0). RTO < 15 minutes. HIPAA (US) and GDPR (EU) compliance. **QUESTION:** How should you design the resource hierarchy and billing to meet the CFO's requirement for cost attribution per hospital tenant?

A new developer joins your team and needs to view the logs for a specific Cloud Run service to troubleshoot an issue. They should not be able to modify the service or view logs for other services. Which IAM role should you grant?

Your company hosts a public-facing web application on GKE behind a Global External HTTP(S) Load Balancer. You notice a sudden spike in traffic originating from a specific country, which is causing performance degradation. The traffic appears to be a Layer 7 DDoS attack. How can you quickly mitigate this?

You have implemented VPC Service Controls to protect your BigQuery datasets. A data scientist working from home needs to query the data using the BigQuery web UI. Their home IP address is dynamic. How can you grant them access without compromising the security perimeter?

Your company requires that all data stored in Cloud Storage be encrypted using keys managed by your security team. The security team wants to automatically rotate the keys every 90 days. Which encryption method should you use?

A media company serves terabytes of video content daily to users globally. They are currently serving the videos directly from Cloud Storage buckets in the US, resulting in high network egress costs and high latency for users in Asia and Europe. Which TWO architectural changes should you implement to reduce costs and improve performance? (Select TWO)

CASE STUDY: ShopGlobal Overview: Retailer, 2000 employees, $500M revenue. US-Central co-lo, Java/Tomcat monolith, Oracle RAC 20TB, batch inventory sync. Business Req: Handle 10x Black Friday spikes, personalized recommendations, modernize to microservices. Execs: CEO wants omnichannel; CFO needs predictable spend; CTO demands zero downtime cutover. Tech Req: PCI-DSS compliance, automated image processing, real-time inventory, CI/CD. Constraints: Complex Oracle stored procedures, team learning containers, strict bi-annual audits. QUESTION: To satisfy the CFO's requirement for predictable cloud spend while handling the 10x Black Friday spikes, what pricing strategy should you use for compute resources?

CASE STUDY: MediSecure Overview: Telehealth provider, 1500 employees, $300M revenue. Core app on AWS, 3 acquired clinics on VMware, fragmented EHRs, Active Directory. Business Req: Unify patient records, integrate clinics in 90 days, launch patient portal. Execs: CEO wants rapid integration; CFO wants CapEx to OpEx; CISO demands strict HIPAA/GDPR compliance. Tech Req: End-to-end PHI encryption, comprehensive audit logging, hybrid connectivity to clinics, DR (RPO 5m, RTO 1h). Constraints: Clinics have low bandwidth, high staff turnover requires automated IAM, legacy EHRs cannot be modified immediately. QUESTION: To ensure the reliability of the new patient portal, the operations team wants to implement SRE practices. How should they define when it is safe to release new features?

Your SRE team is defining Service Level Indicators (SLIs) for a user-facing web application. Which metric is the most appropriate SLI to measure the user experience?

The CFO wants to implement a chargeback model where each department is billed exactly for the cloud resources they consume. Currently, all departments deploy resources into a single shared GCP project. How should you architect the billing solution?

You are running a batch processing job on Compute Engine that takes 4 hours to complete. The job is fault-tolerant; if a VM shuts down, the job can resume from where it left off. You want to minimize compute costs. Which instance type should you use?

Your global e-commerce site serves heavy image and video content. Users in Asia are experiencing slow load times because the origin servers are in the US. You want to improve performance and reduce egress network costs. What should you do?

Your company's GCP bill has increased significantly. You suspect that developers are provisioning VMs that are much larger than necessary. Which TWO tools can help you identify and fix this issue? (Select TWO)

Your company is analyzing its GCP bill and notices that 'Network Egress' is the highest cost. The traffic is primarily users downloading large media files, and internal data transfers between GCP and the on-premises data center. Which THREE strategies will reduce these network costs? (Select THREE)

Your team manages a large Google Kubernetes Engine (GKE) Standard cluster. The CFO has asked you to optimize the compute costs of the cluster. The workloads consist of a mix of critical, always-on web services and fault-tolerant, interruptible batch processing jobs. Which TWO strategies should you implement? (Select TWO)

CASE STUDY: TechStream Gaming Overview: Industry: Gaming Size: 500 employees, $100M revenue Environment: - On-prem US/EU - 200 servers - MySQL (5 TB) - 2M peak users - $100K/mo cost Requirements: - Reduce costs 40% - 5x growth - Launch APAC/SA/Africa - Daily deployments Exec Statements: - CEO: Scale rapidly. - CFO: Max $100K/mo, 18mo ROI. - CTO: Limited cloud exp, 99.95% uptime. Tech Reqs: - <100ms latency globally - Real-time analytics - 5x seasonal spikes - EU data residency - DDoS protection - CI/CD Constraints: - 12mo migration - <4hr downtime - 20 Java/MySQL devs, 5 ops - $2M budget QUESTION: To meet the CFO's requirement of reducing costs by 40%, which pricing strategy should you apply to the baseline, predictable compute workloads?

CASE STUDY: ManuIoT Overview: Industry: Manufacturing Size: 100 factories globally Environment: - 100,000 sensors - Local SCADA - Fragmented SQL Server DBs - No central analytics Requirements: - Predictive maintenance - Real-time global dashboards - Edge computing Exec Statements: - CEO: Monetize telemetry. - CFO: Costs must scale linearly. - VP Ops: Factory lines need local control if internet drops. Tech Reqs: - Ingest 1M msgs/sec - Stream processing - Offline factory capabilities - Train ML centrally, deploy to edge Constraints: - Low bandwidth/high latency at factories - Legacy MQTT protocol - Zero IT staff at factories QUESTION: To meet the requirement of training ML models centrally and deploying them to the edge, which GCP AI service should you utilize?

Your Site Reliability Engineering (SRE) team has defined a Service Level Objective (SLO) of 99.9% availability for a critical API. Over the last 30 days, the API has experienced several outages, and the error budget has been completely exhausted. According to SRE best practices, what action should the team take?

You want to monitor the CPU utilization of a fleet of Compute Engine instances. If the average CPU utilization exceeds 80% for more than 5 minutes, you want the operations team to receive a notification in their Slack channel. How should you configure this?

A startup is running a stateless web application on Compute Engine. The traffic is highly unpredictable, with sudden spikes and long periods of low usage. They want to minimize compute costs but cannot commit to a 1-year or 3-year contract because their architecture might change next month. Which cost optimization strategy is most appropriate?

A media company serves terabytes of video content globally. They are currently hosting the videos in a Cloud Storage bucket in the `us-east1` region. Users in Asia are experiencing high latency, and the company's monthly network egress costs are exorbitant. How should you optimize this architecture?

You are defining Service Level Indicators (SLIs) for a user-facing e-commerce website. According to Google's SRE practices, which TWO metrics are the most appropriate SLIs for a request-driven web service? (Select TWO)

A company is running a large fleet of Compute Engine instances for a batch processing workload. The workload is fault-tolerant, can be interrupted, and takes about 4 hours to complete. They want to drastically reduce their compute costs. Which THREE strategies should they combine to achieve the lowest possible cost? (Select THREE)

Your data analytics team is running complex SQL queries on a 100 TB BigQuery table. The queries are scanning the entire table every time, resulting in massive costs. The queries typically filter data by a `transaction_date` column and group results by a `country_code` column. Which TWO BigQuery features should you implement to reduce the amount of data scanned and lower costs? (Select TWO)

A global enterprise has applications deployed in GCP and on-premises. They are analyzing their monthly GCP bill and notice high network egress costs. Which TWO architectural changes would effectively reduce network egress costs? (Select TWO)

Your company generates 1 TB of log data daily. The data must be immediately accessible for analysis for the first 30 days. From day 31 to 365, it is rarely accessed but must be kept for compliance. After 1 year, it must be kept indefinitely for legal reasons but will almost never be read. Which TWO actions should you take to optimize Cloud Storage costs? (Select TWO)

CASE STUDY: TechStream Gaming Company Overview: TechStream Gaming is a global multiplayer game developer with 500 employees and $100M annual revenue. They recently launched a hit mobile game that is growing rapidly. Current Technical Environment: - On-premises data centers in US and EU. - 200 bare-metal servers running Linux. - Self-managed MySQL databases (5 TB total) for player profiles and inventory. - Peak concurrent users: 2 million. - Current monthly infrastructure cost: $150K. Business Requirements: - Reduce infrastructure costs by 40%. - Support 5x user growth over 2 years. - Launch in 3 new regions (APAC, SA, Africa). - Improve deployment speed from 1 week to daily. Executive Statements: - CEO: "We need to scale rapidly to compete. Cloud migration is critical." - CFO: "Cost reduction is paramount. We cannot exceed $100K/month. ROI must be achieved in 18 months." - CTO: "Our team has limited cloud experience. Reliability is non-negotiable - 99.95% uptime minimum." Technical Requirements: - Sub-100ms latency for players globally. - Real-time analytics on player behavior. - Seasonal traffic spikes (5x during holidays). - CI/CD pipeline for daily deployments. Constraints: - Migration must complete in 12 months. - Cannot exceed 4-hour downtime during cutover. - Dev team: 20 engineers (Java, MySQL). - Ops team: 5 engineers (limited cloud experience). QUESTION: To meet the CFO's cost reduction requirements, which pricing strategy should you implement for the game server compute resources?

CASE STUDY: AutoMakers Inc Company Overview: AutoMakers Inc is a leading vehicle manufacturer transitioning to connected and autonomous vehicles. They need a platform to ingest, process, and analyze telemetry data from millions of cars. Current Technical Environment: - Legacy MQTT brokers on-premises. - Hadoop cluster for batch processing (nightly runs). - 100,000 connected cars sending 1 KB of data every minute. - On-premises data warehouse reaching capacity. Business Requirements: - Support 5 million connected cars within 3 years. - Enable real-time alerting for critical vehicle faults. - Provide predictive maintenance insights to customers. - Monetize anonymized traffic data. Executive Statements: - CEO: "Data is our new engine. We need real-time insights to improve safety." - CFO: "The platform must scale cost-effectively. We only want to pay for what we use." - CTO: "We need a fully managed serverless data pipeline to minimize operational overhead." Technical Requirements: - Ingest up to 1 million messages per second with low latency. - Process data in real-time for anomaly detection. - Store raw telemetry data indefinitely for machine learning model training. - Provide a scalable data warehouse for business intelligence analysts. Constraints: - Strict data privacy regulations (GDPR) require masking of PII. - Limited data engineering staff; prefer managed services. - Must integrate with existing on-premises identity provider (Active Directory). QUESTION: To provide a scalable data warehouse for business intelligence analysts, how should you configure BigQuery to optimize query performance and costs for time-series telemetry data?

You are managing a customer-facing web application deployed on Cloud Run. The development team has created a new version of the application with a redesigned UI. The product manager wants to test the new UI with 10% of the live user traffic to ensure it doesn't negatively impact conversion rates before rolling it out to everyone. How should you implement this?

You are defining the monitoring strategy for a new e-commerce checkout service. You need to establish a Service Level Indicator (SLI) that accurately reflects the user's experience of the service's reliability. Which metric is the most appropriate SLI?

A large enterprise uses a single Shared VPC across 100 different projects. The finance team needs to implement a strict chargeback model to bill individual departments for their specific network egress costs and Compute Engine usage. What is the most reliable way to allocate these costs?

You are advising a traditional IT operations team that is transitioning to a Site Reliability Engineering (SRE) model. They are currently overwhelmed by manual alerts and spend most of their time firefighting. Which TWO SRE practices should you recommend they implement first to reduce toil and improve reliability? (Select TWO)

A global streaming service is planning for a major live event that is expected to generate 10x their normal traffic. They use Compute Engine, Cloud SQL, and Cloud Load Balancing. To ensure capacity and reliability during the event, which THREE actions should they take in advance? (Select THREE)

Your data analytics team uses BigQuery extensively. The CFO has noticed that BigQuery costs have tripled in the last two months. Upon investigation, you find that analysts are running `SELECT *` queries on massive, multi-terabyte tables. Which TWO actions should you take to optimize costs without stopping the analysts from doing their work? (Select TWO)