The central building block of secure and privacy-preserving Vehicular Communication (VC) systems is a Vehicular Public Key Infrastructure (VPKI), which provides vehicles with multiple anonymized credentials, termed pseudonyms. These pseudonyms are used to ensure VC message authenticity and integrity while preserving vehicle (thus passenger) privacy. In the light of emerging large-scale multi-domain VC environments, the efficiency of the VPKI and, more broadly, its scalability are paramount. By the same token, preventing misuse of the credentials, in particular, Sybil-based misbehavior, and managing "honest-but-curious" VPKI entities are other facets of a challenging problem. In this paper, we leverage the state-of-the-art VPKI system and enhance its functionality towards a highly-available, dynamically-scalable, and resilient design; this ensures that the system remains operational in the presence of benign failures or resource depletion attacks, and that it dynamically scales out, or possibly scales in, according to request arrival rates. Our full-blown implementation on the Google Cloud Platform shows that deploying large-scale and efficient VPKI can be cost-effective: the processing latency to issue 100 pseudonyms is approximately 56 ms. More so, our experiments show that our VPKI system dynamically scales out or scales in according to the rate of pseudonyms requests. We formally assess the achieved security and privacy properties for the credential acquisition process. Overall, our scheme is a comprehensive solution that complements standards and can catalyze the deployment of secure and privacy-protecting VC systems.