Hi Michael,
I have made changes to the query and the code used but it seems the query
time for jdbc driver is still slower comapred to Java Rest Binding,
MERGE (firstNode:Soap {id:{1}})
ON CREATE SET firstNode.brand={2}, firstNode.updated = timestamp()
ON MATCH SET firstNode.updated=timestamp()
MERGE (secondNode:Company {id:'{3}'})
ON CREATE SET secondNode.name="{4}", secondNode.updated = timestamp()
ON MATCH SET secondNode.updated=timestamp()
WITH firstNode, secondNode
OPTIONAL MATCH firstNode - [existing:`manufacturer`] - ()
DELETE existing
MERGE (firstNode)-[r:`manufacturer`]-(secondNode);
public void processList(List<CYPHERExecuteQueryObject> list) {
for (CYPHERExecuteQueryObject c : list) {
neo4jjdbcinsert(c.getQuery(), c.getParams());
ctr++;
if (ctr == 500) {
commit();
ctr = 0;
}
}
}
// set auto commit to false
public void neo4jjdbcinsert(String query, Object... params) {
try {
if (con.isClosed()) {
con =
DriverManager.getConnection("jdbc:neo4j://localhost:7474/");
con.setAutoCommit(false);
}
try {
PreparedStatement preparedStatement =
con.prepareStatement(query);
preparedStatement.clearParameters();
for (int i = 0; i < params.length; i++) {
preparedStatement.setObject(i+1, params[i]);
}
preparedStatement.execute();
} catch (Exception e) {
e.printStackTrace();
}
} catch (SQLException e) {
e.printStackTrace();
}
}
public void commit() {
try {
con.commit();
} catch (SQLException e) {
e.printStackTrace();
} finally {
try {
con.close();
} catch (SQLException e) {
e.printStackTrace();
}
}
}
On Tuesday, May 6, 2014 2:37:17 PM UTC+8, Michael Hunger wrote:
>
> 1. You still don't use labels and indexes in MERGE, it should look like
> this:
>
> MERGE (firstNode:Soap {id:{1}})
> ON CREATE SET firstNode.brand={2}
>
> SET firstNode.updated = timestamp()
> 2. You don't use the API like that, but instead create a single
> PreparedStatement then add parameters and execute, then clear, add new
> parameters and execute
> You don't operate on Neo4jConnection but only with the JDBC APIs.
>
> Also this doesn't work: firstNode.brand="{2}"
> it must be: firstNode.brand={2}
>
>
> Am 06.05.2014 um 08:19 schrieb Gene Tan <[email protected] <javascript:>
> >:
>
> Hi Michael.
>
> I tried out using jdbc driver using the same scenario mentioned earlier
> inserting data by batch of 500 queries
> and query
> MERGE (firstNode {id:{1}})
> ON CREATE SET firstNode.brand="{2}", firstNode.updated = timestamp(),firstNode
> :Soap
> ON MATCH SET firstNode.updated=timestamp()
> MERGE (secondNode{id:'{3}'})
> ON CREATE SET secondNode.name="{4}", secondNode.updated = timestamp(),
> secondNode:Company
> ON MATCH SET secondNode.updated=timestamp()
> WITH firstNode, secondNode
> //CHANGE MANUFACTURER if there is an existing manufacturer
> OPTIONAL MATCH firstNode - [existing:`manufacturer`] - ()
> DELETE existing
> CREATE UNIQUE p = (firstNode)-[r:`manufacturer`]-(secondNode) RETURN p;
> and was wondering why jdbc driver processing time is slower that the java
> rest binding, or I am just using the wrong way... because for a small
> sample data 619 nodes and 4611 relationships it took the jdbc driver one
> minute to insert the data while, it just took 21 seconds for the java rest
> binding to insert the data.
> Here is the code used for inserting data:
> public void processList(List<CYPHERExecuteQueryObject> list) {
> for (CYPHERExecuteQueryObject c : list) {
> neo4jjdbcinsert(c.getQuery(), c.getParams());
> ctr++;
> if (ctr == 500) {
> commit();
> ctr = 0;
> }
> }
> }
>
> // set auto commit to false
> public void neo4jjdbcinsert(String query, Object... params) {
> try {
> if (con.isClosed()) {
> con = (Neo4jConnection)
> DriverManager.getConnection("jdbc:neo4j://localhost:7474/");
> con.setAutoCommit(false);
> }
> HashMap<String, Object> map = new HashMap<>();
> for (int i = 0; i < params.length; i++) {
> map.put(i + 1 +"", params[i]);
> }
> try {
> con.executeQuery(query, map);
> } catch (Exception e) {
> e.printStackTrace();
> }
> } catch (SQLException e) {
> e.printStackTrace();
> }
> }
>
> public void commit() {
> try {
> con.commit();
> } catch (SQLException e) {
> e.printStackTrace();
> } finally {
> try {
> con.close();
> } catch (SQLException e) {
> e.printStackTrace();
> }
> }
> }
>
>
> Thanks
>
> On Friday, May 2, 2014 4:14:30 PM UTC+8, Gene Tan wrote:
>>
>> Thanks! Michael, will try this out
>>
>> On Thursday, May 1, 2014 5:15:12 PM UTC+8, Michael Hunger wrote:
>>>
>>> 1. Use labels + indexes (or constraints)
>>>
>>> Without an label + existing index your first operation (merge) will have
>>> to go over all nodes in the graph to find if there is already a duplicate
>>> which it would then merge to
>>> the second CREATE UNIQUE should be a MERGE too.
>>>
>>> 2. Use parameters for your literal input values, otherwise cypher has to
>>> recompile the query every time and cannot reuse-the pre-compiled query
>>> plans.
>>> 3. you might want to move from Java-Rest-Binding to the JDBC driver,
>>> which is much better for cypher interactions with the server.
>>>
>>>
>>>
>>> On Wed, Apr 30, 2014 at 9:25 AM, Gene Tan <[email protected]> wrote:
>>>
>>>> Hello,
>>>>
>>>> I am new to neo4j
>>>> currently already have a graph that contains at least 180000 nodes and
>>>> 1400000
>>>> relationships ,
>>>> I am inserting data by batch of 500 queries, through the java rest
>>>> binding library, and have observed that some queries had encountered read
>>>> timeout, I was wondering if it is because of my query used for inserting
>>>> data. Or is it related to configurations with neo4j..
>>>> Here is a sample query that I am using for inserting data
>>>>
>>>>> MERGE (firstNode {id:'1234'})
>>>>> ON CREATE SET firstNode.brand="Dove", firstNode.updated =
>>>>> timestamp(),firstNode
>>>>> :Soap
>>>>> ON MATCH SET firstNode.updated=timestamp()
>>>>> MERGE (secondNode{id:'2345'})
>>>>> ON CREATE SET secondNode.name="Dove Manufacturer", secondNode.updated
>>>>> = timestamp(), firstNode:Company
>>>>> ON MATCH SET secondNode.updated=timestamp()
>>>>> WITH firstNode, secondNode
>>>>> //CHANGE MANUFACTURER if there is an existing manufacturer
>>>>> OPTIONAL MATCH firstNode - [existing:`manufacturer`] - ()
>>>>> DELETE existing
>>>>> CREATE UNIQUE p = (firstNode)-[r:`manufacturer`]-(secondNode) RETURN p
>>>>> ;
>>>>>
>>>>
>>>> Is there anyway to make this query run faster?
>>>>
>>>> Thanks!
>>>>
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "Neo4j" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to [email protected].
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>>
> --
> You received this message because you are subscribed to the Google Groups
> "Neo4j" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected] <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
>
--
You received this message because you are subscribed to the Google Groups
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
