Question 1

Processing drone telemetry, keeping data rarely accessed → Managed, streaming pipeline → Dataflow + Nearline

🟩 Ingest with ClearBlade IoT Core → Pub/Sub → Dataflow → store in Nearline Cloud Storage
❌ Ingest with IoT Core → store in BigQuery
❌ IoT Core → Dataprep → Coldline
❌ IoT Core → Pub/Sub → BigQuery → Standard storage

Explanation:
You need managed services, real-time processing, and cheap long-term storage.
Nearline is ideal for data accessed monthly.

---

Question 2

KPI for SLO must be measurable and time-bound (latency or success rate).

🟩 Low latency for >85% of requests when aggregated over 1 minute
❌ 85% success aggregated over 1 min (not tied to UX)
❌ 85% success (not time-bound)
❌ 85% satisfied users (subjective)

Explanation:
Latency is the most important SLO metric for user experience.

---

Question 3

Application needs 30 GB RAM → cheapest is a custom machine → not preemptible.

🟩 Stop VM → set-machine-type → 2-custom-4-30720 → Start VM
❌ Use e2-standard-8 + preemptible
❌ Custom + preemptible
❌ e2-standard-8

Explanation:
You need 30 GB RAM + lowest cost → Custom machine.
Preemptible instances are NOT appropriate for production workloads.

---

Question 4

You want custom IP ranges with RFC 1918 class B (172.16.0.0/12).

🟩 Create new project → delete default VPC → create custom VPC → use 172.16.x.x
❌ Auto mode VPC
❌ Leave default VPC
❌ Use 192.168.x.x + peering

Explanation:
To avoid conflicts for future Interconnect → always use custom mode + your own private CIDR planning.

---

Question 5

Two different GCP organizations → need private, low-latency connectivity → VPC Peering allowed across orgs.

🟩 Ensure no IP overlap → enable VPC Network Peering
❌ Same IP ranges
❌ Carrier Peering
❌ Shared VPC (same org required)

Explanation:
Shared VPC does NOT work across different organizations.
VPC Peering enables private, low latency communication without egress charges.

---

Question 6

Your network cannot meet Google’s peering requirements.
You only need access to Google Workspace and Google APIs.

🟩 Use Carrier Peering
❌ Partner Interconnect
❌ Dedicated Interconnect
❌ Direct Peering

Explanation:
Carrier Peering is for public Google services like Workspace, YouTube, Maps, and does NOT require special conditions.

---

Question 7

Need managed NoSQL, low-latency, petabyte-scale → Bigtable

🟩 Create Bigtable → migrate MongoDB → import into Bigtable
❌ Firestore Native
❌ Firestore Datastore
❌ BigQuery

Explanation:
Bigtable is a wide-column, massively scalable, low-latency NoSQL managed database.

---

Question 8

Legacy VM app stores lots of filesystem data → needs stable identity + persistent storage.

🟩 Create a StatefulSet
❌ Container build
❌ Deployment
❌ Single Pod

Explanation:
StatefulSets provide persistent volume claims, ordered startup, and stable network IDs.

---

Question 9

Migrate VMs → containers → automated pipeline.

🟩 Use Migrate for Anthos
❌ Manual GKE + Anthos
❌ Cloud Build
❌ Migrate for Compute Engine

Explanation:
Migrate for Anthos = VM → container migration automation.

---

Question 10

Monolith VM, need to scale-out instead of scaling-up.

🟩 Use containers + GKE autoscaling deployment
❌ Move code + attach volumes
❌ MIG + startup script
❌ VMs + persistent volumes

Explanation:
GKE horizontal pod autoscaling allows scale-out.

---

🧾 Summary Table

---