Graph Traversal Workload

read heavy

Summary

Multi-hop relationship traversal through connected entities where depth and branching factor drive query cost. Access patterns are read-heavy with point lookups to seed traversals. Query planning must bound traversal depth to prevent exponential scan growth.

Example Systems

·Fraud ring detection (connected accounts with common attributes)
·Social graph friend-of-friend recommendation
·Access control permission inheritance via role hierarchy
·Knowledge graph querying for entity relationships
·Supply chain dependency tracing
·Co-purchase recommendation graph

Characteristics

CategoryMIXED

Read / write patternread heavy

Latency requirementlow

Consistency requirementsession

Durability requiredYes

Ordering requiredNo

Capacity

Typical RPS500

Peak RPS5,000

Typical data volume100 GB

Growth rate5-30 GB/month; node and edge counts grow with user activity

Access Patterns

graph traversalpoint lookup

Recommended Patterns

cache asidematerialized viewread replica

Patterns to Avoid

sharding

Basis

Graph workload characteristics are specific and well-documented; traversal depth bounding is a standard operational requirement in graph database literature