Connection pooling for VM protocol?

[Paul Gale <pa...@gmail.com>]

Hi,

Is connection pooling needed when sending to an _embedded_ ActiveMQ broker
from Camel using the VM protocol vm://? If so, why?

According to page 288 of the Camel In Action book the VM protocol bypasses
the TCP stack tends to make me think not.

The example given here http://camel.apache.org/activemq.html outlines how
to implement pooling to a remote broker over TCP which makes the need for
the pooling clear.

Before someone mentions it, yes, I know this question straddles the worlds
of ActiveMQ and Camel but I have to pick one of the lists to post to.

Thanks,
Paul

Re: Fwd: Connection pooling for VM protocol?

[artnaseef <ar...@artnaseef.com>]

Connection pooling is a solution to the problem of overhead in creating and
discarding connections when using a technology that doesn't support
long-lived connections.

If the technology used is straight JMS, there should be no need to use
pooling.

What client technology is it?




--
View this message in context: http://activemq.2283324.n4.nabble.com/Fwd-Connection-pooling-for-VM-protocol-tp4680442p4680482.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Re: Connection pooling for VM protocol?

[Gary Tully <ga...@gmail.com>]

the broker maintains some per connection state, connectionId and
clientId to enforce the single clientId in use jms requirement, this
can be an area of thread contention on heavily loaded systems that
open/close many connections.
So even for vm connections (where there is no tcp socket) there will
be a pooling benefit.

But the age old answer is to 'ask the computer' for your particular use case.

On 22 April 2014 17:07, Paul Gale <pa...@gmail.com> wrote:
> Hi,
>
> Connection pooling is generally recommended as a good thing. However, is
> pooling of VM protocol (vm://) based connections
> required/advised/nonsensical when communicating between Camel and a broker
> in which it is embedded? Please explain.
>
> From what I have read the VM protocol bypasses the TCP stack, therefore I'm
> inclined to think it's not necessary. Regardless, that doesn't necessarily
> imply that there isn't a cost associated with creating and destroying these
> vm:// connections such that it wouldn't benefit from pooling.
>
> Thanks,
> Paul



-- 
http://redhat.com
http://blog.garytully.com

Fwd: Connection pooling for VM protocol?

[Paul Gale <pa...@gmail.com>]

Hi,

Connection pooling is generally recommended as a good thing. However, is
pooling of VM protocol (vm://) based connections
required/advised/nonsensical when communicating between Camel and a broker
in which it is embedded? Please explain.

>From what I have read the VM protocol bypasses the TCP stack, therefore I'm
inclined to think it's not necessary. Regardless, that doesn't necessarily
imply that there isn't a cost associated with creating and destroying these
vm:// connections such that it wouldn't benefit from pooling.

Thanks,
Paul