You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@karaf.apache.org by hsinchen <hc...@gmail.com> on 2015/08/27 09:39:40 UTC
Remote OSGi Performance Compared to Same JVM or Two JVM
For a consumer bundle to access the provider bundle within a JVM, the
performance should be better than the two standalone (consumer and provider)
applications deployed in two JVMs with SOAP or REST. With DOSGi (Remote
OSGi) by CXF, it goes with SOAP or REST, too. Then, the performance is the
same just like the two standalone applications with SOAP or REST. Is it
right? If yes, any of ways to get better performance near the same JVM.
--
View this message in context: http://karaf.922171.n3.nabble.com/Remote-OSGi-Performance-Compared-to-Same-JVM-or-Two-JVM-tp4042077.html
Sent from the Karaf - User mailing list archive at Nabble.com.
Re: Remote OSGi Performance Compared to Same JVM or Two JVM
Posted by hsinchen <hc...@gmail.com>.
Does the "OSGi Remote Services/RSA specification" mean "Remote OSGi" or
"Distributed OSG"?
The scenario is that we usually use SOAP or REST for a consumer and a
provider in each standalone VM (JVM) as case 0. In OSGi, the consumer and
the provider may be in the same VM (JVM) as case 1, in the different VM but
a same Cellar cluster as case 2, or in the different VM without Cellar
cluster as case 3.
Normally, the perform of case 1 is better than case 2 and case 3. I would
like to know if the Remote OSGi spec. handles this interaction between the
consumer and provider in case 2 or case 3 with the better performance just
better than SOAP or REST in pure JEE world (case 0).
--
View this message in context: http://karaf.922171.n3.nabble.com/Remote-OSGi-Performance-Compared-to-Same-JVM-or-Two-JVM-tp4042077p4042107.html
Sent from the Karaf - User mailing list archive at Nabble.com.
Re: Remote OSGi Performance Compared to Same JVM or Two JVM
Posted by Scott Lewis <sl...@composent.com>.
On 8/27/2015 12:39 AM, hsinchen wrote:
> For a consumer bundle to access the provider bundle within a JVM, the
> performance should be better than the two standalone (consumer and provider)
> applications deployed in two JVMs with SOAP or REST. With DOSGi (Remote
> OSGi) by CXF, it goes with SOAP or REST, too. Then, the performance is the
> same just like the two standalone applications with SOAP or REST. Is it
> right? If yes, any of ways to get better performance near the same JVM.
One way to go is to create your app using the OSGi Remote Services/RSA
specification, and use an appropriate distribution provider (or even
create your own provider if your needs dictate). As an example, ECF's
RSA implementation currently has the following distribution providers:
1) tcp/socket-based (generic)
2) socket/websocket-based (rosgi)
3) JMS (all protocols)
4) Jax-RS
5) MQTT
6) Hazelcast
7) Custom
Each of these have different performance characteristics, based upon
what transports are being used, but all of them implement the OSGi R6
RSA specification. Another advantage of such an approach is the ability
to use different distribution systems for development, testing, and
deployment (perhaps into different environments) without modifying the
application.
Scott
[1] https://wiki.eclipse.org/ECF
Re: Remote OSGi Performance Compared to Same JVM or Two JVM
Posted by Jean-Baptiste Onofré <jb...@nanthrax.net>.
Cellar DOSGi does the same as CXF DOSGi without the overhead of
SOAP/REST (directly using hazelcast NIO/serialization).
In the same JVM, you can increase the throughput with multiple
consumer/thread on the same service. It's not really better performance,
but more throughput.
Regards
JB
On 08/27/2015 09:39 AM, hsinchen wrote:
> For a consumer bundle to access the provider bundle within a JVM, the
> performance should be better than the two standalone (consumer and provider)
> applications deployed in two JVMs with SOAP or REST. With DOSGi (Remote
> OSGi) by CXF, it goes with SOAP or REST, too. Then, the performance is the
> same just like the two standalone applications with SOAP or REST. Is it
> right? If yes, any of ways to get better performance near the same JVM.
>
>
>
>
>
> --
> View this message in context: http://karaf.922171.n3.nabble.com/Remote-OSGi-Performance-Compared-to-Same-JVM-or-Two-JVM-tp4042077.html
> Sent from the Karaf - User mailing list archive at Nabble.com.
>
--
Jean-Baptiste Onofré
jbonofre@apache.org
http://blog.nanthrax.net
Talend - http://www.talend.com