cancel
Showing results for 
Search instead for 
Did you mean: 

Head's Up! These forums are read-only. All users and content have migrated. Please join us at community.neo4j.com.

Storage size

Hi. I am developing an ontology in Neo4j Desktop. I have a few quetions:

  1. I try to find the 'actual' data storage size. Firstly I think I need to remove the 'logical log' type data, and what else shall I exclude?
    2X_e_e026945087192ff58b1e9133e4d17b2e29fa0083.png 2X_8_8ede8e73e3b212db45e75288a399e8a9f17de5a5.png
    As you can see the sample case, I deleted all the 'acted in' type relationships, and some of the data types like 'lable store', 'relationship store' are almost double. Is this reasonable?
  2. I want to build a similar ontology model in Protege, but it seems much smaller in storage size, which is different from what I learn from literature. Is that because of Marginal utility?

Thank you!

1 REPLY 1

Though you do not indicate with what version of Neo4j this has been experienced on

  1. :sysinfo and 'Logical Log:' size in 'Store Sizes' represents the size of the logical log currently being written to. So if you have set up your transaction log retention/rotation https://neo4j.com/docs/operations-manual/3.5/configuration/transaction-logs/ to be rotate at 250mb and keep 10 transaction logs, then the 'Logical log:' size will report a number less than 250mb, even though you may have 9 other 250mb transaction logs.

  2. the doubling in size is as a result of how deletes and reuse works. For example if you add 1m nodes this will consume data in data/databases/graph.db/neostore.nodestore.db. If you then delete those 1m nodes we update those 1m nodes in neostore.nodestore.db and set each with a IN-USE flag of false, thus no real net change to size difference, AND then copy each of the internal ids for these 1m nodes to neostore.nodestore.db.id, which thus has a increase in size. This .id file effectively serves as a recycle bin such that if you were then to add back 1m nodes we would drain the neostore.nodestore.db.id of 1m ids, thus reducing its size to near 0, and re-use those ids.