Tuesday, 19 June 2018

Thoughts on graph database

  1. Could one say, that a graph database is a relational database but stores the references as physical addresses instead of logical connections that have to be translated to physical addresses on retrieval? Good for 1:1 but what if 1:n or n:m? Probably not single reference address stored but linked list.
  2. If 1 holds true, is data stored on disk like that or is there a translation between memory and disk?
  3. Are graph databases bidirectional by priciple?
  4. If bidirectional, are there storage penalties?
  5. Performance advantage is especially good with queries involving more than one reference, in relational terms involving more than one join. Maybe this is considered in the spot light of 6
  6. Performance advantage is especially good when starting at one spot and retrieving related data to this spot. For set operations it seems to be the other way round.
  7. Apparently, graph databases store redundant data when they are said to use more storage.
  8. There cannot be, by principle, a link/reference without both ends of the link.
References:

No comments:

Post a Comment

[git] Create a local branch from another branch

From the active branch git checkout -b <LOCAL_BRANCH> From a donator branch git checkout -b <LOCAL_BRANCH> <DONATING_BRANCH...