Possible representations
Flat node list
- Can partially load models -> central point to intercept for missing nodes
- Properties are stored directly in node
- Disadvantage: Also have to go through list for child relations
-> All nodes/edges are equal
- Might align nicely with (global) index
- Works the same way for graph DBs
- Possibly more performant than tree, as it can be chunked easily
Tree
- Easier to traverse than a graph
Graph
- Fits inherent structure
- Disadvantage: No obvious unique serialization
=> Decision: use flat node list
Possible representations
Flat node list
-> All nodes/edges are equal
Tree
Graph
=> Decision: use flat node list