You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@qpid.apache.org by "Rajith Attapattu (JIRA)" <qp...@incubator.apache.org> on 2008/05/29 01:16:45 UTC
[jira] Created: (QPID-1102) A new java.io based blocking transport
for client
A new java.io based blocking transport for client
-------------------------------------------------
Key: QPID-1102
URL: https://issues.apache.org/jira/browse/QPID-1102
Project: Qpid
Issue Type: New Feature
Components: C++ Client
Affects Versions: M3
Reporter: Rajith Attapattu
Assignee: Rajith Attapattu
Fix For: M3
Attachments: transport.patch
I have experimented with a java.io based blocking transport which gives a dramatic improvement in latency and memory while maintaining the same or much better throughput rates than MINA. Also from testing I see that MINA is using more CPU.
I have tested it extensively with 10 hour soak tests and short duration perf tests using sonic harness.
In order to use the new transport you need to specify -Dtransport=io.
MINA still remains the default transport.
If the number of connections are small and if you can devote a thread per connection, then you the blocking transport has an advantage over nio bcos it eleminates the overhead of a select/poll or epoll system call.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (QPID-1102) A new java.io based blocking transport
for client
Posted by "Aidan Skinner (JIRA)" <qp...@incubator.apache.org>.
[ https://issues.apache.org/jira/browse/QPID-1102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Aidan Skinner resolved QPID-1102.
---------------------------------
Resolution: Fixed
This is the default fo 0-10 now
> A new java.io based blocking transport for client
> -------------------------------------------------
>
> Key: QPID-1102
> URL: https://issues.apache.org/jira/browse/QPID-1102
> Project: Qpid
> Issue Type: New Feature
> Components: Java Client
> Affects Versions: M3
> Reporter: Rajith Attapattu
> Assignee: Rajith Attapattu
> Fix For: M3
>
> Attachments: client.patch, transport.patch
>
>
> I have experimented with a java.io based blocking transport which gives a dramatic improvement in latency and memory while maintaining the same or much better throughput rates than MINA. Also from testing I see that MINA is using more CPU.
> I have tested it extensively with 10 hour soak tests and short duration perf tests using sonic harness.
> In order to use the new transport you need to specify -Dtransport=io.
> MINA still remains the default transport.
> If the number of connections are small and if you can devote a thread per connection, then you the blocking transport has an advantage over nio bcos it eleminates the overhead of a select/poll or epoll system call.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (QPID-1102) A new java.io based blocking transport
for client
Posted by "Rajith Attapattu (JIRA)" <qp...@incubator.apache.org>.
[ https://issues.apache.org/jira/browse/QPID-1102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Rajith Attapattu updated QPID-1102:
-----------------------------------
Component/s: (was: C++ Client)
Java Client
changed the component to java client
> A new java.io based blocking transport for client
> -------------------------------------------------
>
> Key: QPID-1102
> URL: https://issues.apache.org/jira/browse/QPID-1102
> Project: Qpid
> Issue Type: New Feature
> Components: Java Client
> Affects Versions: M3
> Reporter: Rajith Attapattu
> Assignee: Rajith Attapattu
> Fix For: M3
>
> Attachments: client.patch, transport.patch
>
>
> I have experimented with a java.io based blocking transport which gives a dramatic improvement in latency and memory while maintaining the same or much better throughput rates than MINA. Also from testing I see that MINA is using more CPU.
> I have tested it extensively with 10 hour soak tests and short duration perf tests using sonic harness.
> In order to use the new transport you need to specify -Dtransport=io.
> MINA still remains the default transport.
> If the number of connections are small and if you can devote a thread per connection, then you the blocking transport has an advantage over nio bcos it eleminates the overhead of a select/poll or epoll system call.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (QPID-1102) A new java.io based blocking transport
for client
Posted by "Rajith Attapattu (JIRA)" <qp...@incubator.apache.org>.
[ https://issues.apache.org/jira/browse/QPID-1102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Rajith Attapattu updated QPID-1102:
-----------------------------------
Attachment: transport.patch
> A new java.io based blocking transport for client
> -------------------------------------------------
>
> Key: QPID-1102
> URL: https://issues.apache.org/jira/browse/QPID-1102
> Project: Qpid
> Issue Type: New Feature
> Components: C++ Client
> Affects Versions: M3
> Reporter: Rajith Attapattu
> Assignee: Rajith Attapattu
> Fix For: M3
>
> Attachments: transport.patch
>
>
> I have experimented with a java.io based blocking transport which gives a dramatic improvement in latency and memory while maintaining the same or much better throughput rates than MINA. Also from testing I see that MINA is using more CPU.
> I have tested it extensively with 10 hour soak tests and short duration perf tests using sonic harness.
> In order to use the new transport you need to specify -Dtransport=io.
> MINA still remains the default transport.
> If the number of connections are small and if you can devote a thread per connection, then you the blocking transport has an advantage over nio bcos it eleminates the overhead of a select/poll or epoll system call.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (QPID-1102) A new java.io based blocking transport
for client
Posted by "Rajith Attapattu (JIRA)" <qp...@incubator.apache.org>.
[ https://issues.apache.org/jira/browse/QPID-1102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Rajith Attapattu updated QPID-1102:
-----------------------------------
Attachment: client.patch
> A new java.io based blocking transport for client
> -------------------------------------------------
>
> Key: QPID-1102
> URL: https://issues.apache.org/jira/browse/QPID-1102
> Project: Qpid
> Issue Type: New Feature
> Components: C++ Client
> Affects Versions: M3
> Reporter: Rajith Attapattu
> Assignee: Rajith Attapattu
> Fix For: M3
>
> Attachments: client.patch, transport.patch
>
>
> I have experimented with a java.io based blocking transport which gives a dramatic improvement in latency and memory while maintaining the same or much better throughput rates than MINA. Also from testing I see that MINA is using more CPU.
> I have tested it extensively with 10 hour soak tests and short duration perf tests using sonic harness.
> In order to use the new transport you need to specify -Dtransport=io.
> MINA still remains the default transport.
> If the number of connections are small and if you can devote a thread per connection, then you the blocking transport has an advantage over nio bcos it eleminates the overhead of a select/poll or epoll system call.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.