Microservices scaling: independent vs coordinated deployment strategies

In theory, microservices promote independent scaling and deployment. You can scale your user service without touching your product catalog. In practice, however, we often find ourselves needing to coordinate deployments and scaling actions across multiple services. For example, if we're doing a major feature release that touches our API gateway, authentication service, and user profile service, scaling one often implies scaling the others, or at least ensuring they can handle the increased load or new data contracts. This 'co-ordinated scaling' can feel like an anti-pattern to the microservices philosophy. Are we doing something wrong, or is this just the practical reality of complex systems with interdependent services? It sometimes feels like we've traded a monolithic scaling problem for a distributed scaling coordination problem. How do you balance the ideal of independent scaling with the practical necessities of a tightly integrated system?