A multi-tenant analytics platform uses AKS for two workloads: (1) an always-on API tier with steady traffic and (2) large, nightly batch jobs that can tolerate restarts. Azure SQL Database and an App Service-based admin portal run 24x7 and have predictable usage. Leadership mandates a 30% reduction in compute cost without impacting SLOs. Which two recommendations best meet the goal?

Adding an AKS spot node pool for nightly batch jobs is cost-effective because spot instances offer significant discounts, and batch jobs are fault-tolerant, making them suitable for preemption. Using PodDisruptionBudgets/tolerations and checkpointing ensures resilience against preemption, maintaining SLOs. Purchasing 1-year reserved capacity for Azure SQL Database and App Service provides substantial cost savings (up to 72%) for predictable, always-on workloads. Scaling AKS to zero during business hours is incorrect as the API tier needs to be always on. Migrating to B-series VMs is not ideal for an API tier with steady traffic as B-series are designed for burstable workloads and might not meet performance SLOs. Disabling autoscale would prevent scaling up during peak demand, potentially impacting performance and availability, thus violating SLOs.