On 14/03/2008, Carl Trieloff <[EMAIL PROTECTED]> wrote:

> sure, as you know qpid get's included into fedora, this feature has been
> picked up to also
> transport OS info, CPU, interrupts, network etc.
>
> Another project that I know that has picked it up is looking to push
> data around hypervior info
>
> And yet another is doing work laying sec alerts, and resource
> utilization over the same mechanism. It turns
> out it is very easy to create another mgmt object and integrate it.

Interesting. This raises some further questions in my mind:

1) What applications/tools view and process the data? i.e. if a group
puts CPU info into the management messages, what do they read it in?

2) For some of these it sounds like the message format is being used
as a general format for management info that is pretty much
independent of the broker (or Qpid). Is this intended to replace or
compete with WS-M?

> not yet ... but multiple parties in AMQP Working Group have expressed
> support for it, one
> of the rabbit users has been working with it also - I am sure hit will
> get updated somehow before
> making it's way into a release. Discussion so far is in the WG is that
> it would be optional to impl as part
> of the spec.

The advantage of this I suppose is that if multiple broker
implementations supported it, users could manage their installations
from a single tool. Or, more realistically, if users wrote scripts
against the JMX interface, those would be portable across different
brokers.

I think it would be useful to know how concrete the Rabbit (or other
vendor) plans are.

> yes agree, thus I was suggesting that it is nice project for someone to
> pick up, so the
> bridge exists and then makes it valuable for Java broker when it gets to
> it. The questions
> is -- if we do a bridge as a project would that also be interesting for
> the Java broker down the road?

I think it would be interesting. Even in the short term, it would be
useful for the C++ and Java brokers to share the management verbs and
properties. From a user perspective if the bridge supported WS-M and
JMX then the C++ (via the bridge) and Java brokers (natively) would in
theory support the same set of tools.

> yes, there are two places we can map this, at the broker into the
> AMQP-mgmt protocol or
> at the impl of the bridge. I think the former is the better approach --
> please comment if you
> don't agree

I think that AMQP-mgmt is a better approach if it gets proper buy-in.

RG

Reply via email to