Anyway to have multiple databases on a neo4j instance?

Question

Coming from a relational database mindset, it seems odd that one only one one graph db per instance of neo4j. Is the idea that we do multiple subgraphs starting from root ?

Answer 1

If it is only for development, you can use neo4j-instance, it helps me work on several neo4j databases at once (at the same time, basically setting each on a different port), and it flows a lot better than adding a label to differentiate between different schema, or prefixing nodes with the scheme name. Since, you do not have to change how you interact with the neo4j databases because they are still separate databases on different ports.

Answer 2

The concept of "root" node is going away. There are many problems with this, most of which revolve around node density.

I believe the heart of your question is around database design, and whether it is smarter to have several graph database instances, or one instance with several subgraphs.

Really it's up to you, but I would go with the subgraph idea as it allows some of your data to be shared in the same connection, and Neo4j doesn't really any performance penalties if you do this, provided you keep them separated, then the only problem you'll eventually run into is the max size of nodes/relationship, but this is an artificial limit that will be bumped up later.

Answer 3

Yes , I have recently created multiple neo4j instance in pytest session level fixture , suppose you want to run multiple neo4j instance in localhost, here assuming host is single (localhost), and if you see the neo4j conf file($neo4j_home/conf/neo4j.conf) , default configurations are for http port 7474 and for bolt 7687 , Suppose one neo4j is running with http port 7474 , now you can not run another neo4j instance with 7474 , you have to change the port in neo4j.conf then you can start another instance , to resolve this

sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.bind(("", 0))
sock.listen(1)
port = sock.getsockname()[1]

It will give you the free port and need to change this in neo4j.conf in runtime now lets see this pytest fixture

@pytest.fixture(scope="session")
def neo4j_graph_instance(csv_node_file, csv_relation_file):
    instancesDirectory = os.path.join(PATH_TEST_DIR, "neo4j_instance_acceptance")
if not os.path.exists(instancesDirectory):
    os.makedirs(instancesDirectory)

archive_file = "neo4j-community-3.2.1-unix.tar.gz"
filename = instancesDirectory + "/" + archive_file

if (os.path.isfile(filename)) == True:
    try:
        with TarFile.open(filename, "r:gz") as archive:
            archive.extractall(instancesDirectory)
            print ("successfully extracted")

    except IOError:
        print "file is not there"
else:
    try:
        uri = "http://dist.neo4j.org/neo4j-community-3.2.1-unix.tar.gz"
        urlretrieve(uri, filename)
        try:
            with TarFile.open(filename, "r:gz") as archive:
                archive.extractall(instancesDirectory)
        except IOError:
            print "file is not there"
    except IOError:
        print "Could not connect to internet to download the file"

neo4j_home = os.path.join(instancesDirectory, "neo4j-community-3.2.1")
neo4j_import_dir = os.path.join(neo4j_home, "import")
neo4j_inst = Neo4jInstance(home=neo4j_home)

neo4j_inst.set_config("dbms.connector.http.listen_address", ":%s" % get_open_port())
neo4j_inst.set_config("dbms.connector.bolt.listen_address", ":%s" % get_open_port())
neo4j_inst.set_config("dbms.security.auth_enabled", "false")
neo4j_inst.set_config("dbms.connector.https.enabled", "false")

# db loading mechanism #
# Rajib: I am getting csv files fixture and copying them in neo4j import dir  in run time.
# Then getting bolt uri for that active instance  , and executing the command 'load_db_script'.
# At the end I am returning one Graph instance with fully loaded with csv files.


neo4j_inst.start_neo4j_instance()
time.sleep(15)
#:todo need to avoid sleep , instead check socket connection to desired host and port
print "copying csv files to neo4j import direcotry"

copy_script = "cp %s %s %s" % (
    csv_node_file, csv_relation_file,  neo4j_import_dir)
call(copy_script, shell=True)
print "successfully copied to neo4j import directory "
neo4j_inst_bolt_uri = "bolt://localhost:%d" % neo4j_inst.get_bolt_port()
load_cypher_file = os.path.join(DMS_PATH, "load_csv.cypher")
load_db_script = "cat %s | %s -a %s --format verbose --debug" % (
    load_cypher_file, neo4j_inst.cypher_shell_script(), neo4j_inst_bolt_uri)
call(load_db_script, shell=True)
neo4j_inst_http_port = neo4j_inst.get_http_port()

yield Graph(host='localhost', http_port=neo4j_inst_http_port, bolt=False)

print "running teardown methods"

neo4j_inst.stop_neo4j_instance()
neo4j_inst.delete_store()

So every time it will give you a fixture with neo4j instance where port will be assigned in run time with fully loaded database.