On 21 Jun 2009, at 05:04, Macarse wrote: > I gave up my count(*) test and try something that hurts relational > databases!
I think your example is still one which actually matches relational databases quite nicely. Try this one as an example: You have a social networking site, in which people can be friends with each other; in SQL, you'd have one 'people' table and one many-to-many join table with two foreign keys to the people table, representing a 'friendship' between two people. Now you could have several goals: - like LinkedIn, you want to show how many people are friends, friends- of-friends and friends-of-friends-of-friends of a particular user (without counting someone twice) - you want to find the shortest chain of links which connects two given users of the system (a kind of "six degrees of separation" experiment: http://en.wikipedia.org/wiki/Six_degrees_of_separation ) You cannot do these with a single SQL statement, but you have to iterate queries. That's where relational databases really start hurting. If you cannot get this example to be faster in Neo4j than in MySQL, I will be rather disappointed of Neo. Cheers Martin _______________________________________________ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user