On 3/30/07, David Powell <David.Powell at sun.com> wrote: > A good reason for doing this is because you want to take advantage of > the relationship between the service and the instance. The primary > element of this relationship is the fact that instances inherit > properties from services, and you can have multiple instances per > service. > > One possible application of this split is a service which defines all > the defaults and creates instances as-needed to fulfill some need. > For example, svc:/network/ssl/proxy defines a set of defaults for a > Kernel SSL service, and ksslcfg dynamically creates instances when > new SSL servers are desired. No SSL server is needed out-of-the-box, > so the service ships with no instances.
In my site, we ran a couple of ldap servers. Each instance has some settings that are local but for most part, they are the same. We basically abstracted everything that we can about the instance into the service definition so we only need to do minimal definition on the instance front. In fact, since our start/stop script recognized the instance SMF supplies to it, the instance definition only defines the name of the instance. -- Just me, Wire ...
