Eric,
My James instance is pretty lite; it has only the SMTP server, no
user/mailbox, no JMX. For using the KahaDB datastore I changed the
<amq:persistenceAdapter> section in james-server.xml, replacing the
<amq:amqPersistenceAdapter/> line with <amq:kahaDB
directory="filesystem=file://var/store/activemq/brokers/KahaDB"
journalMaxFileLength="64mb" indexCacheSize="50000"/>. I also replaced
the activeio-core, activemq-core and kahadb jars with the ones from
the ActiveMQ-5.6.0 distribution.
Numbers bellow. I only measured the inbound pipeline, so the "tps"
value represent how many messages were enqueued per second. The
numbers in the square brackets tell how many concurrent threads were
hammering the James instance during that test iteration. Each
iteration consisted in 5k requests. Please take them with a grain of
salt.
Hardware where James ran:
Hardware: Xen: VirtualHost
CPU: 1 x Intel(R) Xeon(R) CPU E5507 @ 2.27GHz
RAM: 1900MB
The code that generated the load was run on a different box.
[5]avg=8.551, tps=116.9453865045024, errors=0, msg_size=4kilos
[10]avg=7.8336, tps=127.65522875816994, errors=0, msg_size=4kilos
[15]avg=5.556, tps=179.98560115190784, errors=0, msg_size=4kilos
[20]avg=6.69, tps=149.47683109118086, errors=0, msg_size=4kilos
[25]avg=7.6082, tps=131.43713361899006, errors=0, msg_size=4kilos
[5]avg=10.4, tps=96.15384615384616, errors=0, msg_size=8kilos
[10]avg=7.8588, tps=127.24588995775437, errors=0, msg_size=8kilos
[15]avg=8.4324, tps=118.59019970589631, errors=0, msg_size=8kilos
[20]avg=8.28, tps=120.77294685990339, errors=0, msg_size=8kilos
[25]avg=6.6016, tps=151.47842947164324, errors=0, msg_size=8kilos
[5]avg=17.6774, tps=56.56940500299818, errors=0, msg_size=12kilos
[10]avg=9.8692, tps=101.32533538686013, errors=0, msg_size=12kilos
[15]avg=9.2196, tps=108.46457546965162, errors=0, msg_size=12kilos
[20]avg=8.6836, tps=115.15961122115252, errors=0, msg_size=12kilos
[25]avg=7.4592, tps=134.06263406263406, errors=0, msg_size=12kilos
[5]avg=21.606, tps=46.28343978524484, errors=0, msg_size=16kilos
[10]avg=10.8896, tps=91.83073758448428, errors=0, msg_size=16kilos
[15]avg=9.8158, tps=101.87656635220766, errors=0, msg_size=16kilos
[20]avg=8.7852, tps=113.82780130219004, errors=0, msg_size=16kilos
[25]avg=8.2, tps=121.95121951219512, errors=0, msg_size=16kilos
[5]avg=20.4012, tps=49.01672450640159, errors=0, msg_size=20kilos
[10]avg=11.452, tps=87.32099196646874, errors=0, msg_size=20kilos
[15]avg=10.204, tps=98.00078400627204, errors=0, msg_size=20kilos
[20]avg=9.5742, tps=104.44736897077563, errors=0, msg_size=20kilos
[25]avg=9.4026, tps=106.35356178078405, errors=0, msg_size=20kilos
-Valentin
On Wed, Sep 19, 2012 at 4:35 AM, Eric Charles <[email protected]> wrote:
Hi Valentin,
That's good news!
Any patch or performance figures you could provide?
Thx, Eric
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
