Suppose some Windows service uses code that wants mapped network drives and no UNC paths. How can I make the drive mapping available to the service's session when the service is started? Logging in as the service user and creating a persistent mapping will not establish the mapping in the context of the actual service.
You'll either need to modify the service, or wrap it inside a helper process: apart from session/drive access issues, persistent drive mappings are only restored on an interactive logon, which services typically don't perform.
The helper process approach can be pretty simple: just create a new service that maps the drive and starts the 'real' service. The only things that are not entirely trivial about this are:
The helper service will need to pass on all appropriate SCM commands (start/stop, etc.) to the real service. If the real service accepts custom SCM commands, remember to pass those on as well (I don't expect a service that considers UNC paths exotic to use such commands, though...)
Things may get a bit tricky credential-wise. If the real service runs under a normal user account, you can run the helper service under that account as well, and all should be OK as long as the account has appropriate access to the network share. If the real service will only work when run as LOCALSYSTEM or somesuch, things get more interesting, as it either won't be able to 'see' the network drive at all, or require some credential juggling to get things to work.