RRK Interview Handbook — Google Cloud CE III (L6)

🧠 Interview Frameworks Print & Memorize

Six frameworks that structure EVERY answer. Print the condensed guide and review 10 min before your call.

🚀

DAPMO — Migration

Discover → Assess → Plan → Migrate → Optimize

"Dad Plans Moves Out"

🔄 USE WHEN: Move from on-prem / AWS / Azure to GCP

Discover — Inventory workloads, dependencies, data gravity, compliance needs
Assess — TCO analysis, cloud-readiness scoring, risk identification
Plan — Migration waves, lift-and-shift vs. re-architect, timeline & milestones
Migrate — Execute — Migrate for CE, DMS, Transfer Appliance
Optimize — Right-size, committed use discounts, modernize (containers, serverless)

🏗️

USDVP — Solution Design

Understand → Scope → Design → Validate → Present

"US Delivery VP"

📋 USE WHEN: Customer wants to build X — architect it

Understand — Business outcome, SLA requirements, users, existing stack
Scope — What's in/out, constraints (budget, timeline, compliance)
Design — Components, data flow, GCP services, trade-offs
Validate — PoC / pilot, load test, security review, customer feedback
Present — Communicate to both technical and executive stakeholders

🤝

ACREN — Objection Handling

Acknowledge → Clarify → Reframe → Evidence → Next Step

"ACRES of land"

🗣️ USE WHEN: Customer pushes back / prefers AWS / refuses to adopt

Acknowledge — Validate their concern — "That's a fair point..."
Clarify — Ask what's behind it — technical? political? contractual?
Reframe — Shift from product comparison to business outcome
Evidence — Reference architecture, case study, PoC data, TCO comparison
Next Step — Propose a low-risk pilot or workshop to demonstrate value

📋

RCDSFS — Architecture Whiteboard

Requirements → Components → Data Flow → Scale → Failure → Security

"Really Cool Designs Scale Fast & Securely"

🖍️ USE WHEN: Draw / design the architecture for this system on GCP

Requirements — Clarify users, throughput, latency, storage, compliance
Components — Draw high-level boxes — compute, storage, networking, ML/data
Data Flow — Arrows — how data moves through the system end to end
Scale — Handle 10x traffic? Auto-scaling, CDN, sharding, caching
Failure — What breaks? Multi-region, retries, circuit breakers, DR
Security — IAM, encryption at rest/transit, VPC, DLP, audit logging

👥

ADAEM — Stakeholder Engagement

Align → Discover → Advise → Enable → Measure

"Advising ADAM + Enable"

🤝 USE WHEN: Build a relationship with a new strategic account

Align — Org chart, priorities, decision-makers, buying center
Discover — Tech stack, pain points, cloud maturity, upcoming initiatives
Advise — Map goals to GCP capabilities, propose a phased roadmap
Enable — Workshops, hands-on labs, PoCs, architecture reviews
Measure — Adoption metrics, consumption growth, CSAT, expansion

🔧

IROPT — Troubleshooting

Isolate → Remediate → Optimize → Prevent

"I Repair Old Problems"

⚠️ USE WHEN: Something broke / latency / failures / outages

Isolate — Narrow blast radius — logs, metrics, trace the path, identify failing component
Remediate — Quick fix to restore service — rollback, failover, scale up
Optimize — Root cause the issue, tune the fix for long-term stability
Prevent — Monitoring, alerting, runbooks, chaos testing to avoid recurrence

🎯 Pro Tip: "Let me walk through my approach — first I'd [step 1]..." signals methodical thinking. Lead EVERY answer with the framework name: "Using the RCDSFS framework, let me walk through..."

☁️ Category 1: Cloud Technology Landscape

1 A customer is running a monolithic Java application on-prem. They want to modernize. Walk me through your approach. ▶

Framework: DAPMO (Migration) + USDVP (Solution Design)

📋 Approach:

Discover: Understand the app — dependencies, traffic patterns, team skills, business drivers for modernizing (cost? agility? scale?)
Assess: Evaluate modernization paths — can we containerize as-is? Do we need to decompose into microservices? What's the risk tolerance?
Plan: Phased approach — Phase 1: containerize the monolith (lift-and-containerize to GKE), Phase 2: incrementally extract microservices using the strangler fig pattern
Migrate: Move the containerized app to GKE; use Cloud SQL for the database initially
Optimize: Break out high-value services (e.g., payment, search) into independent microservices; adopt Cloud Run for stateless services; add Pub/Sub for async communication

☁️ GCP Services: GKE (≈ EKS), Cloud Run (≈ Fargate), Cloud SQL (≈ RDS), Pub/Sub (≈ SNS/SQS), Cloud Build (≈ CodePipeline), Artifact Registry (≈ ECR)

🎯 L6 Level-Up: An L6 answer starts with "What's driving the modernization?" — not jumping straight to tech. Show you'd align the approach to their business outcome (faster release cycles? cost? scaling for growth?).

2 What are the key differences between lift-and-shift, re-platform, and re-architect? When would you recommend each? ▶

Framework: USDVP (Solution Design)

📋 Approach:

Lift-and-shift: Move as-is to cloud VMs. Fast, low risk, minimal code changes. Recommend when: tight timeline, legacy app with no active development, or as Phase 1 before modernizing. GCP: Compute Engine + Migrate for Compute Engine
Re-platform: Minor adjustments to leverage managed services. Moderate effort. Recommend when: app is actively maintained, team wants quick wins (managed DB, auto-scaling) without full rewrite. GCP: GKE, Cloud SQL, Memorystore
Re-architect: Redesign for cloud-native. High effort, high reward. Recommend when: app needs to scale 10x, current architecture is a bottleneck, or team is building new features anyway. GCP: Cloud Run, Pub/Sub, Firestore, Spanner

🎯 L6 Level-Up: Don't just list the three — recommend a portfolio approach: lift-and-shift the 80% of apps that are stable, re-platform the 15% with active development, and re-architect the 5% that are strategic differentiators.

3 A customer has workloads on AWS and wants to adopt a multicloud strategy with GCP. How would you approach this? ▶

Framework: ADAEM (Stakeholder Engagement) + USDVP (Solution Design)

📋 Approach:

Align: Understand WHY multicloud — vendor lock-in concerns? Best-of-breed services? Regulatory? M&A brought in different platforms?
Discover: Map their current AWS workloads — what's portable vs. deeply integrated (e.g., heavy Lambda + DynamoDB = harder to move)
Advise: Recommend "best workload for best cloud" — don't move everything; identify GCP strengths that solve specific problems
Enable: Start with a non-critical workload that showcases GCP's differentiation; use Anthos for consistent management across both clouds
Measure: Track adoption, compare cost/performance, expand as confidence builds

☁️ GCP Services: Anthos (no AWS equiv), BigQuery (≈ Redshift), Cloud Interconnect (≈ Direct Connect), GKE Enterprise (≈ EKS Anywhere)

🎯 L6 Level-Up: Don't push "move everything to GCP." Position GCP for where it wins (data/analytics/AI) and build a land-and-expand strategy. Show commercial awareness — start with a winnable workload that demonstrates ROI, then grow.

4 How would you migrate a customer's on-prem data warehouse to BigQuery? What challenges would you anticipate? ▶

Framework: DAPMO (Migration)

📋 Approach:

Discover: Current platform (Teradata? Oracle? SQL Server?), data volume, ETL jobs, number of users/reports, SLAs
Assess: Schema complexity, stored procedures, custom UDFs, legacy SQL dialect differences, data sensitivity/compliance
Plan: Parallel-run strategy — migrate in phases (historical data first, then live feeds)
Migrate: BigQuery Data Transfer Service or Dataflow for ETL; Storage Transfer Service for bulk data; translate stored procs to BigQuery SQL or Dataform
Optimize: Move from reserved slots to on-demand pricing; partitioning/clustering; Looker for reporting

☁️ GCP Services: BigQuery (≈ Redshift), Dataflow (≈ Glue/Kinesis Data Analytics), BigQuery Data Transfer Service, Dataform (≈ dbt), Looker (≈ QuickSight)

🎯 L6 Level-Up: Mention the organizational change management aspect — it's not just a tech migration. Users, report authors, and data teams need enablement. An L6 thinks about people + process, not just technology.

5 A retail customer wants real-time analytics on their sales data. What GCP services would you recommend and why? ▶

Framework: USDVP (Solution Design)

📋 Approach:

Understand: What does "real-time" mean to them — seconds? sub-second? What decisions are they making? (inventory, pricing, fraud?)
Scope: Data sources (POS systems, e-commerce, mobile), volume, who consumes insights?
Design: Ingest via Pub/Sub → Process via Dataflow → Store in BigQuery → Serve via Looker
Validate: Run a PoC with one data source (e.g., top 10 stores) to prove latency meets requirements

☁️ GCP Services: Pub/Sub (≈ Kinesis Data Streams), Dataflow (≈ Kinesis Data Analytics), BigQuery (≈ Redshift), Bigtable (≈ DynamoDB) if low-latency needed, Looker (≈ QuickSight)

🎯 L6 Level-Up: Ask "What action will you take when you see the data?" — this separates a dashboard project from a real-time decisioning system. If they need automated responses (e.g., dynamic pricing), add Cloud Functions triggered by Pub/Sub.

6 How does GKE compare to EKS? A customer on EKS asks why they should consider GKE. ▶

Framework: ACREN (Customer Objection)

📋 Approach:

Acknowledge: "EKS is solid. You've invested in it and it works."
Clarify: "What pain points, if any, do you experience?" (upgrade complexity? node management? multi-cluster visibility?)
Reframe: Position GKE's differentiators against their specific pains:
- Autopilot mode — fully managed node infrastructure. EKS Fargate profiles are less seamless.
- Release channels — automated, safe cluster upgrades. EKS upgrades are more manual.
- Multi-cluster management — GKE Enterprise (Anthos) provides fleet management.
- Integrated security — Binary Authorization, Workload Identity baked in.
- Cost — Autopilot = pay per pod resource, not per node.
Evidence: Reference Google's upstream Kubernetes contributions
Next Step: Propose a side-by-side PoC — run the same workload on both, compare operations burden

🎯 L6 Level-Up: Don't trash EKS. Position GKE as reducing operational burden so their team can focus on app development instead of cluster management. Frame it as "give your developers time back."

7 A customer wants to build an ML pipeline to predict customer churn. What GCP services would you recommend? ▶

Framework: USDVP (Solution Design)

📋 Approach:

Understand: What data do they have? (usage logs, billing, support tickets, demographics) How will predictions be consumed?
Scope: Team ML maturity — do they have data scientists, or need AutoML?
Design the pipeline: Data prep via BigQuery → Training via Vertex AI → Serving via Vertex AI Endpoints → Orchestration via Vertex AI Pipelines
Validate: Start with AutoML as a baseline, then iterate with custom models

☁️ GCP Services: Vertex AI (≈ SageMaker), BigQuery ML (no direct equiv), Dataflow (≈ Glue), Cloud Composer (≈ MWAA), Looker (≈ QuickSight)

🎯 L6 Level-Up: Mention BigQuery ML as a differentiator — "Your analysts can build a churn model with SQL, no Python needed." This is a Google-unique selling point that resonates with customers whose data teams are SQL-first.

8 What is Anthos and when would you propose it to a customer? ▶

Framework: USDVP (Solution Design)

📋 What it is: A managed platform for running Kubernetes workloads consistently across GCP, on-prem, AWS, and Azure — single pane of glass for policy, security, and observability.

✅ When to propose it:

Customer has workloads across multiple clouds and wants unified management
Customer can't fully migrate to cloud (regulatory, data gravity) but wants cloud-native dev experience
Customer wants consistent security policies across all environments
Customer is doing a phased migration and needs to run the same workloads in both places

❌ When NOT to propose it:

Customer is all-in on one cloud — just use GKE directly
Small startup with no multicloud complexity
Customer's problem is better solved by a simpler migration

☁️ GCP Services: Anthos / GKE Enterprise, Anthos Service Mesh (≈ App Mesh), Anthos Config Management (GitOps), Connect Gateway

🎯 L6 Level-Up: Position Anthos as the bridge, not the destination. "We use Anthos to give you a consistent developer experience today while you migrate at your own pace. It de-risks the journey." Don't over-pitch it for simple use cases.

🏗️ Category 2: Solution Architecture — "Design X on GCP"

Use the RCDSFS framework for every whiteboard question: Requirements → Components → Data Flow → Scale → Failure → Security

9 Design YouTube on GCP — video upload, processing, and streaming at scale ▶