You need single-digit millisecond reads on billions of time-series rows by key. Which store fits?
Why B is the answer
Bigtable
Bigtable handles high-throughput, low-latency key lookups. BigQuery is built for analytics.
In detail
- Bigtable gives fast reads by key over billions of rows.
- Design the row key around your reads.
Use case
An illustrative scenario, not a real incident.
A monitoring product reads a device's last hour of metrics by key, within milliseconds, from billions of rows in Bigtable.
See it in different ways
flowchart TD
A{"Access pattern"} -->|"Analytics in SQL"| B["BigQuery"]
A -->|"Huge key lookups"| C["Bigtable"]
A -->|"Global relational consistency"| D["Spanner"]
A -->|"Regional relational app"| E["Cloud SQL or AlloyDB"]
A -->|"Objects and files"| F["Cloud Storage"]
A -->|"In-memory cache"| G["Memorystore"]mindmap
root((Stores))
BigQuery
Analytics
Bigtable
Key lookups at scale
Spanner
Global relational
AlloyDB
PostgreSQL
Firestore
Mobile documents
Memorystore
Microsecond cache
Cloud Storage
Objectsflowchart TD
Q{"How is data read?"}
Q -->|"By key, fast"| A0["Bigtable"]
Q -->|"Analytic SQL"| A1["BigQuery"]Key point: Bigtable serves fast key lookups at scale.
Why the other options are wrong
- A. Cloud SQL
- Cloud SQL will not scale to this.
- C. BigQuery
- BigQuery is for analytics.
- D. Cloud Storage
- Object storage is not for key lookups.