jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Akil Ali <Akil.Aj...@cognizant.com>
Subject Performance test conducted on Jackrabbit
Date Sat, 14 Feb 2009 11:28:33 GMT

Hi,
   I have performed several test on jackrabbit to check the read and write
operation on jackrabbit.

Scenario: I have to add thousands of documents (say MILLION) into the
jackrabbit. so for that purpose i have performed several test to check the
following.
1> How many documents application can add into the jackrabbit/Second.
2>How many documents application can read/Second.

Note: I am not using versioning and also i have disabled indexing of
jackrabbit

1> How many documents application can add into the jackrabbit/Second.

For that i have integrated threading while adding documents.
i have performed the test with a single thread and with multiple thread

Jackrabbit node structure looks like
Root Node that has 2 node A & B.

Single Thread:while adding 50000 documents i have performed 2 iteration. in
each iteration i have added 25000 documents to each node. in iteration 1 i
have added 25000 documents to node A and then in iteration 2 25000
documents.

2 Thread: 
Thread1-->Session1-->Node A---->Adding 25000 documents
Thread2-->Session2-->NodeB--->Adding 25000 documents
i dont synchronized the code for adding documents in jackrabbit.

Please see the attachment excel with this mail to see the time taken to add
the article with single thread and with multiple thread. see sheet "Added"

You can see even if i have used threading the performance of Single thread
is better then multiple thread. WHY the performance of single thread is
better than multiple thread used?

2>How many documents application can read/Second.
Jackrabbit structure looks like

-Repository has 2 parent nodes A and B. 
-Each parent node has 25000 docs. 
Test is performed and 25000 nodes are read randomly.
a> By UUID and 2>By Direct access

Please see the attachment excel with this mail to see the time taken to read
the article By UUID and By Direct access. see sheet "Read Test"

There is only slightly difference between both the operation.
Can you please explain WHY?
http://www.nabble.com/file/p22011732/JCRTest.zip JCRTest.zip 
-- 
View this message in context: http://www.nabble.com/Performance-test-conducted-on-Jackrabbit-tp22011732p22011732.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.


Mime
View raw message