(ref.doc)cr63-31

Designing for persistence
ODBMS
by Henry Strickland
pp 31-35

object creation
ownership (or garbage collection, or reference counting)
 - roots and trees
select vs navigate
traversals
 - locality of reference (objects used together)
 - order of traversal (same for better performance)
clusters and group operations
 - paged architectures
 - locking
client-server
weak database semantics (dirty reads and queries)
failure modes
caching objects
 - association objects (caching containers)
keeping data classes clean
 - separate "data" from "application" classes
exceptions
 - resumption
 - termination
 - trickle-back (return error code: non working with operators)