You are designing a Spark Structured Streaming application to process a large dataset in real-time for a financial analytics platform. The platform requires high throughput and low latency to handle millions of transactions per second. During the initial testing phase, you notice that the performance of your queries is significantly impacted by data serialization overhead. Considering the need for cost efficiency, compliance with financial data regulations, and scalability, which of the following optimizations would BEST address the serialization issue while meeting all the platform's requirements? (Choose one option.)