Dynamic Scaling and Consistency in Cloud Storage Systems: The Anna Framework for Autoscaling, Multi-Tiered Key-Value Storage
Nandagopal Parise · 2025
As cloud storage systems become increasingly complex, managing scalability and consistency while maintaining optimal performance is a significant challenge. The Anna framework offers a novel approach to addressing these challenges by providing an autoscaling, multi-tiered, coordination-free, distributed key-value store service. Built on AWS components, Anna dynamically adapts to varying workload demands through horizontal elasticity, multi-master selective replication, and vertical tiering. These mechanisms enable Anna to adjust its storage capacity and computational resources based on workload volume and access patterns, ensuring cost-performance trade-offs are met. The system’s design prioritizes simplicity and efficiency by avoiding shared-memory coordination and leveraging gossip-+based updates to handle conflicts asynchronously. This paper explores Anna’s architecture, focusing on its dynamic adjustment capabilities, its ability to balance consistency and availability, and its innovative approach to monitoring and enforcing service-level objectives (SLOs) without compromising on performance.