IONA vs Visigenic vs ORL performance

IONA vs Visigenic vs ORL performance

Post by Adam Bubl » Wed, 22 Apr 1998 04:00:00



Hi,
I update the page about the throughput test we performed. Orbix,
VisiBroker and omniORB were tested. The results are from local machine
as well as over the net (10BASE-T Ethernet). There are sources, results
in graphs and some conclusion from the test.

Orbix is the slowest from the evaluated ORBs, but the speed is not the
only criterion. For example MICO (I didn't try it, but I will :) is
about the same speed as Orbix and is widely used.

http://www.kav.cas.cz/~buble/corba/comp/test/through/

Regards,
        Adam

 
 
 

IONA vs Visigenic vs ORL performance

Post by Sai-Lai L » Wed, 22 Apr 1998 04:00:00



> I update the page about the throughput test we performed. Orbix,
> VisiBroker and omniORB were tested. The results are from local machine
> as well as over the net (10BASE-T Ethernet). There are sources, results
> in graphs and some conclusion from the test.

> Orbix is the slowest from the evaluated ORBs, but the speed is not the
> only criterion. For example MICO (I didn't try it, but I will :) is
> about the same speed as Orbix and is widely used.

> http://www.kav.cas.cz/~buble/corba/comp/test/through/

Interesting comparison.

In addition to the baseline measurements, it would be interesting to see a
comparison on how the ORBs scale as the number of concurrent clients
increases. In omniORB, we have been careful to minimise thread
contentions and thread switching within the runtime. This pays off as a low
per-call overhead. I believe this will also have a significant positive
impact as the load on the server increases.

--
Dr. Sai-Lai Lo                          |       Research Scientist
                                        |

                                        |       24a Trumpington Street
Tel:            +44 223 343000          |       Cambridge CB2 1QA
Fax:            +44 223 313542          |       ENGLAND

 
 
 

IONA vs Visigenic vs ORL performance

Post by Prasad Mokkapat » Wed, 22 Apr 1998 04:00:00




> > I update the page about the throughput test we performed. Orbix,
> > VisiBroker and omniORB were tested. The results are from local machine
> > as well as over the net (10BASE-T Ethernet). There are sources, results
> > in graphs and some conclusion from the test.

> > Orbix is the slowest from the evaluated ORBs, but the speed is not the
> > only criterion. For example MICO (I didn't try it, but I will :) is
> > about the same speed as Orbix and is widely used.

> > http://www.kav.cas.cz/~buble/corba/comp/test/through/

> Interesting comparison.

> In addition to the baseline measurements, it would be interesting to see a
> comparison on how the ORBs scale as the number of concurrent clients
> increases. In omniORB, we have been careful to minimise thread
> contentions and thread switching within the runtime. This pays off as a low
> per-call overhead. I believe this will also have a significant positive
> impact as the load on the server increases.

Another interesting comparisons would be to look at how the ORB
scales when a multi-threaded clients makes simultaneous calls on
to the server from different threads.
 
 
 

IONA vs Visigenic vs ORL performance

Post by Sai-Lai L » Wed, 22 Apr 1998 04:00:00



> Another interesting comparisons would be to look at how the ORB
> scales when a multi-threaded clients makes simultaneous calls on
> to the server from different threads.

Agree.

--
Dr. Sai-Lai Lo                          |       Research Scientist
                                        |

                                        |       24a Trumpington Street
Tel:            +44 223 343000          |       Cambridge CB2 1QA
Fax:            +44 223 313542          |       ENGLAND