Soru

Zorluk: OrtaPlanning Database Options and Persistent Disk Types

An online retail organization is migrating its platform backend to Google Cloud. The architecture requires two storage components: first, a transactional relational database for order management that mandates ACID compliance and automatic cross-zone failover within a single region; second, ultra-low sub-millisecond latency scratch storage for ephemeral data processing jobs running on Compute Engine instances. Which TWO storage and database configurations should the team select to satisfy these requirements?

  1. Cloud SQL configured with High Availability (regional failover) for the order management databaseCevap
  2. Local SSD attached to Compute Engine instances for high-performance temporary scratch storageCevap
  3. C
    Cloud Bigtable configured with multi-cluster replication for the order management database
  4. D
    Cloud Storage Coldline storage bucket mounted to VM instances for high-performance temporary scratch storage

Cevap

The team should select Cloud SQL configured with High Availability for the transactional relational order management database, and Local SSD attached to Compute Engine instances for low-latency temporary scratch storage.
Cloud SQL with High Availability (HA) provides managed relational engines with ACID compliance and automatic failover between primary and standby zones in a region. Local SSDs are physically attached to the server hosting the Compute Engine VM, offering high throughput and sub-millisecond latency required for temporary scratch processing.

Adım Adım Çözüm

1
Analyze the relational transactional database requirements
Identify that Cloud SQL provides managed SQL engines (MySQL, PostgreSQL, SQL Server) with ACID compliance, and configuring High Availability adds automatic cross-zone failover within a region.
Relational transactional workloads requiring ACID compliance and regional redundancy align directly with Cloud SQL HA.
2
Analyze the high-performance scratch storage requirements
Identify Local SSD as host-attached ephemeral block storage providing maximum IOPS and sub-millisecond latency.
Temporary/scratch workloads requiring sub-millisecond response times are best served by host-attached Local SSD.
3
Evaluate wrong options against constraints
Cloud Bigtable lacks relational multi-table ACID transactions, and Coldline Cloud Storage carries high retrieval fees and higher latency meant for archival data.
Selecting NoSQL for relational SQL needs or cold object storage for active scratch space violates GCP architectural best practices.

Anahtar Kavram

Matching GCP relational databases (Cloud SQL HA) and block storage types (Local SSD) to specific operational and performance requirements
Bu soruyu puanla