Skip to content

Latest commit

 

History

History
16 lines (13 loc) · 960 Bytes

why.md

File metadata and controls

16 lines (13 loc) · 960 Bytes

Why develop another "graph" / "relation" storage

TL;DR: easy setup, simple scaling, maintainability, heavy write throughput, HTTP API, ttl feature

In detail

  • with billions of edges and thousands of gigabytes stored common graph databases struggle to deliver sub millisecond response times for node and edge history requests
  • graph databases like neo4j have do not give access to their scaling features in the community versions
  • graph databases like janusgraph have require complicated setups
  • graph databases like orientDB or grakn do not perform well under heavy write throughput, especially when creating thousands of edges per second
  • graph databases like titan or FlockDB are not maintained anymore
  • most existing graph databases miss "exists" features that result in the execution of multiple queries
  • graph databases like neo4j do not have a ttl feature, nodes and edges cannot expire automatically