You are designing a Service Fabric application for a trading platform that maintains user session state with sub-millisecond access and must survive node failures. Upgrades must be performed with zero downtime and automatic rollback if health degrades. Which two design choices should you make?

Implementing stateful services with Reliable Collections is correct because Service Fabric's Reliable Collections provide highly available, low-latency (sub-millisecond) in-cluster persistence for state, ensuring session state survives node failures. Configuring rolling upgrades with health monitoring and automatic rollback across upgrade domains is correct because this strategy ensures zero downtime during upgrades and automatically reverts to a stable state if issues arise, meeting the requirement for resilient upgrades. Choosing Bronze reliability with a 3-node cluster is incorrect because Bronze reliability offers minimal guarantees and a 3-node cluster provides limited fault tolerance, which is insufficient for a critical trading platform requiring high availability. Specifying Gold reliability and deploying at least seven nodes is incorrect because while Gold reliability and more nodes increase resilience, it's not a direct answer to the core requirements of state persistence and upgrade strategy. Performing cluster-wide upgrades in one batch is incorrect as this would cause downtime and doesn't allow for health monitoring or automatic rollback, failing to meet the zero-downtime and resilient upgrade requirements.