Examkin

Question

Difficulty: Very hardSelecting and Designing Storage and Database Resources

A connected-vehicle enterprise is designing a global telemetry platform to ingest real-time sensor metrics from millions of vehicles worldwide. The architecture must handle over 2×1062 \times 10^6 write operations per second with single-digit millisecond write latency, store semi-structured time-series data with dynamic schemas, and allow data analysts to execute ad-hoc SQL queries directly through BigQuery without creating periodic ETL pipelines. Which storage and database architecture should you recommend?

  1. Deploy Cloud Bigtable with multi-cluster replication across targeted regions for low-latency time-series ingestion, and query the data directly from BigQuery using external table definitions.Answer
  2. B
    Deploy Cloud Spanner with a multi-region instance configuration, using strongly consistent relational tables to index high-velocity time-series sensor streams.
  3. C
    Collect telemetry logs locally on edge gateway servers and continuously dispatch Transfer Appliance hardware to upload data into Cloud Storage Coldline buckets for analytical querying.
  4. D
    Deploy Cloud SQL for PostgreSQL with horizontal read replicas, relying exclusively on IAM roles without VPC Service Controls to govern database access and prevent data export.

Answer

Deploy Cloud Bigtable with multi-cluster replication across targeted regions for low-latency time-series ingestion, and query the data directly from BigQuery using external table definitions.
Cloud Bigtable provides scalable NoSQL key-value storage optimized for massive throughput time-series ingestion with single-digit millisecond latency. Integrating BigQuery via external table definitions enables direct SQL analytics without maintaining complex streaming ETL pipelines.

Step-by-Step Solution

1
Analyze workload write throughput and data model requirements.
The requirement specifies over 2×1062 \times 10^6 write operations per second with single-digit millisecond latency and semi-structured time-series data. Cloud Bigtable is the optimal NoSQL key-value store designed specifically for high-volume time-series ingestion.
Relational databases like Cloud SQL or Spanner are inefficient and cost-prohibitive for raw high-velocity key-value sensor append streams.
2
Evaluate analytical query requirements without ETL overhead.
BigQuery supports direct querying of Cloud Bigtable via external table federated queries, allowing analysts to run SQL without needing custom ETL streaming pipelines.
Federated querying preserves operational performance while fulfilling ad-hoc analytical access requirements.
3
Verify architectural fit against constraints.
Cloud Bigtable multi-cluster replication provides regional resiliency and localized low latency while integrating seamlessly with BigQuery for analytics.
Meets all performance, storage model, cost, and operational requirements.

Key Concept

Selecting Cloud Bigtable for high-throughput time-series ingestion combined with BigQuery federated queries for zero-ETL analytics.
Rate this question

Topics

Selecting and Designing Storage and Database Resources