Examkin

Soru

Zorluk: OrtaDesigning Infrastructure for Technical Requirements and High Availability

A healthcare diagnostics company is building a high-availability infrastructure on Google Cloud to ingest and process high-resolution medical imaging scans from on-premises diagnostic centers. The sustained peak bandwidth requirement between the on-premises environment and Google Cloud is 8 Gbps8\text{ Gbps} with a mandatory 99.99%99.99\% availability Service Level Objective (SLO). The backend image processing pipeline runs memory-intensive image parsing workloads on Compute Engine virtual machines. Which TWO architectural solutions should you implement to satisfy both the high-availability hybrid networking and auto-scaling technical requirements? (Select TWO)

  1. Establish redundant Cloud Dedicated Interconnect connections across two distinct edge availability domains for 99.99% availability.Cevap
  2. Configure Compute Engine Managed Instance Group (MIG) autoscaling based on custom Cloud Monitoring metrics measuring memory utilization.Cevap
  3. C
    Provision an HA VPN gateway with two active IPsec tunnels to handle the sustained 8 Gbps bandwidth requirement.
  4. D
    Configure Compute Engine Managed Instance Group autoscaling based strictly on an 80% target average CPU utilization metric.
  5. E
    Use VPC Network Peering to transit traffic from the on-premises network through a central transit VPC into spoke workload VPCs.

Cevap

The correct architectural choices are establishing redundant Dedicated Interconnect connections across two edge availability domains and configuring Compute Engine Managed Instance Group autoscaling using custom Cloud Monitoring memory utilization metrics.
Dedicated Interconnect satisfies high-bandwidth demands (8 Gbps8\text{ Gbps}) that exceed single IPsec VPN tunnel limits, and deploying across dual edge availability domains delivers a 99.99%99.99\% availability SLA. Furthermore, memory-bound compute workloads require scaling rules tied to actual memory pressure via Cloud Monitoring custom metrics rather than CPU utilization.

Adım Adım Çözüm

1
Evaluate hybrid bandwidth and availability requirements
Sustained bandwidth of 8 Gbps8\text{ Gbps} exceeds HA VPN single-tunnel limits (3 Gbps3\text{ Gbps} per tunnel). Dedicated Interconnect with links across two edge availability domains is required to support high throughput and achieve a 99.99%99.99\% uptime SLA.
HA VPN cannot natively scale to 8 Gbps on a single tunnel, whereas Dedicated Interconnect provides 10 Gbps or 100 Gbps pipes with 99.99% availability when deployed redundantly across edge domains.
2
Identify resource bottlenecks for compute autoscaling
The medical imaging parsing workload is memory-intensive rather than CPU-bound.
Autoscaling on CPU utilization will cause memory exhaustion errors before CPU thresholds trigger. Using Cloud Monitoring custom metrics for memory consumption ensures accurate auto-scaling behavior under load.

Anahtar Kavram

Designing High Availability Hybrid Networking and Metric-Specific Compute Scaling
Tahmini Süre:2m 0s
Bu soruyu puanla

Konular

Designing Infrastructure for Technical Requirements and High Availability