On 02/03/2010 02:35 PM, Darryl L. Pierce wrote: > On Wed, Feb 03, 2010 at 01:50:37PM -0500, Perry Myers wrote: >> This is a good start to genericizing things. One thing to keep in mind is >> that it is better to use some sort of standard external interface for >> doing as much as possible rather than providing infinite and arbitrary >> hooks/scripts plugins. >> >> i.e. rather than providing N hooks for allowing remote configuration of a >> node for stateless mode, instead provide a single interface using a well >> known standard like AMQP/REST/dnssrv to do this. Then other users of the >> node can just adopt the standard and move forward without needing to >> modify the node image itself. > > I agree that working with a standard is the way to go, and using any > combination of the above is better. But how about data encoding; i.e., > how to package together the hardware description, for example? The above > describe a few different ways of discovering the management server, but > don't provide a way to encode data to be delivered. Are there any standards > for that? I'm not aware of any, personally.
/me shudders, but CIM probably provides a standard for this. But in the absence of this, as long as we provide a well formed API and keep it consistent as we improve functionality that is a start. i.e. this is how libvirt evolved. Perry _______________________________________________ Ovirt-devel mailing list [email protected] https://www.redhat.com/mailman/listinfo/ovirt-devel
