Hi Isuru, Reka

In the json format of the service group we define a group name, which gives 
each group a unique identifier when it is deployed. Since the grouping we are 
currently adding . / implementing allows to auto scale entire groups (and 
nested groups) we have to make sure that we are uniquely identify group 
instances of a deployed group (by means of a unique group instance Id in 
addition to the group name). For example, if a group of name A is auto scaled, 
the system will create multiple group instances of group A but each instance 
separately is distinguished by a unique identifier (e.g. A, 1 / A, 2 / A, 3).

Also, I think it will be necessary to add the group name and instance Id to the 
relevant events of the VMs created as part of the group, e.g. Member Created 
Event, Member Terminated Event, etc … . This will allow any interested party 
(e.g. topology listener) to keep track of VMs belonging to a group instance.

Btw are we planning to retain the current dynamic (as implemented in the POC) 
group object model or do you expect changes to it ?

Thanks

Martin


Reply via email to