If a single service is reused by multiple business processes, then the relationships between the service and the multiple processes becomes a factor in performance monitoring and management. For example, sub-optimizing individual service performance will not help a poor process that calls the service 50 times before sending a response to the user, and these 50 requests creat a backlog from the three other processes that also use that service.
This means that performance monitoring is no longer "collect some red-yellow-green metrics and display them in a GUI." Instead performance monitoring is "analyze the metrics in the context of the service relationships in the production environment and display that information in different ways for different audiences." The different audiences being anyone who has a stake in managing performance -- developers, business process architects, operations staff, business managers, etc.
Similarly, companies can no longer rely on "throwing more resources at it" to improve performance. This means that performance management is evolving into a multidisciplinary activity. Managing performance means being able to investigate a wide range of problem causes beyond resource bottlenecks, such as process architecture, software changes, code quality, and changes in service usage. So, be prepared for more collaboration and look for tools that can simplify the contextual analysis and share the results.