Question

Difficulty: Very hardDesigning Infrastructure for Technical Requirements and High Availability

An international financial institution is designing a core transaction platform on Google Cloud. The application requires a relational database capable of global multi-region active-active processing, strong transactional consistency, zero Recovery Point Objective (RPO), and an RTO of under 5 seconds during a full regional outage. Additionally, audit operations require dedicated, SLA-backed private network connectivity from an on-premises data center to Google Cloud capable of sustaining 10 Gbps dedicated bandwidth without traversing the public internet. Which architecture fulfills these high availability and connectivity requirements while adhering to Google Cloud best practices?

  1. Deploy a multi-region Cloud Spanner instance across two active regions with a witness region, and establish redundant Dedicated Interconnect connections with Private Google Access enabled for private endpoint communication.Answer
  2. B
    Deploy a Cloud SQL for PostgreSQL instance configured with high availability (HA) in the primary region and an asynchronous cross-region read replica, combined with redundant Dedicated Interconnect connections.
  3. C
    Deploy a multi-region Cloud Spanner instance, and establish an HA VPN setup with multiple IPsec tunnels to carry the 10 Gbps on-premises hybrid traffic.
  4. D
    Deploy a multi-region Cloud Spanner instance in a central database VPC, and use VPC Network Peering to transitively route on-premises traffic from an edge hybrid VPC into the central database VPC.

Answer

Deploy a multi-region Cloud Spanner instance across two active regions with a witness region, combined with redundant Dedicated Interconnect connections utilizing Private Google Access.
Cloud Spanner multi-region deployment uses synchronous replication across regions via Paxos consensus, delivering global ACID compliance, zero RPO, and automatic failover with sub-second RTO. Pairing this with Dedicated Interconnect ensures private, SLA-backed 10 Gbps physical network throughput for on-premises hybrid traffic via Private Google Access.

Step-by-Step Solution

1
Analyze storage and transactional availability requirements.
Identified that multi-region active-active relational transactions with zero RPO and sub-second RTO necessitate Cloud Spanner rather than Cloud SQL.
Cloud SQL only supports regional HA with synchronous replication within a single region; cross-region Cloud SQL replication is asynchronous and incurs data loss risk (RPO > 0).
2
Analyze hybrid connectivity throughput and SLA requirements.
Identified that SLA-backed 10 Gbps dedicated bandwidth requires Dedicated Interconnect.
Cloud VPN operates over the public internet and maxes out at 3 Gbps per tunnel, whereas Dedicated Interconnect provides 10 Gbps or 100 Gbps private physical pipes directly to Google.
3
Evaluate network routing topology.
Confirmed Private Google Access over Interconnect allows on-premises systems to query Cloud Spanner endpoints privately without relying on invalid transitive VPC Peering configurations.
VPC Network Peering is non-transitive and cannot forward on-premises traffic across peered VPC boundaries.

Key Concept

Multi-region High Availability and Dedicated Hybrid Networking
Rate this question