OceanStore is a global-scale persistent storage infrastructure designed to provide continuous access to data across an untrusted server network. It aims to address the challenges of ubiquitous computing by ensuring data security, durability, and availability. Key features include: 1. **Data Protection**: Data is protected through redundancy and cryptographic techniques. 2. **Caching**: Data can be cached anywhere, anytime, enhancing performance. 3. **Adaptation**: Monitoring usage patterns allows for adaptation to regional outages and denial-of-service attacks. 4. **Performance**: Proactive movement of data optimizes performance. 5. **Scale**: The system supports over $10^{13}$ files, serving approximately $10^{10}$ users with at least 10,000 files each. The system is built on a distributed, fault-tolerant architecture with a focus on deep archival storage, where data is encoded using erasure codes and spread across hundreds or thousands of servers. This ensures that information can survive major disasters and regional outages. OceanStore also supports flexible update mechanisms, conflict resolution, and a robust API for application developers. The paper outlines the system's architecture, including naming, access control, data location, and update models. It discusses the use of probabilistic and global algorithms for data location, and the implementation of conflict resolution through a primary tier of replicas and a secondary tier of replicas. The system's deep archival storage mechanism uses erasure codes to ensure data durability, and the API provides full access to OceanStore functionality, enabling applications to interact with the system effectively.OceanStore is a global-scale persistent storage infrastructure designed to provide continuous access to data across an untrusted server network. It aims to address the challenges of ubiquitous computing by ensuring data security, durability, and availability. Key features include: 1. **Data Protection**: Data is protected through redundancy and cryptographic techniques. 2. **Caching**: Data can be cached anywhere, anytime, enhancing performance. 3. **Adaptation**: Monitoring usage patterns allows for adaptation to regional outages and denial-of-service attacks. 4. **Performance**: Proactive movement of data optimizes performance. 5. **Scale**: The system supports over $10^{13}$ files, serving approximately $10^{10}$ users with at least 10,000 files each. The system is built on a distributed, fault-tolerant architecture with a focus on deep archival storage, where data is encoded using erasure codes and spread across hundreds or thousands of servers. This ensures that information can survive major disasters and regional outages. OceanStore also supports flexible update mechanisms, conflict resolution, and a robust API for application developers. The paper outlines the system's architecture, including naming, access control, data location, and update models. It discusses the use of probabilistic and global algorithms for data location, and the implementation of conflict resolution through a primary tier of replicas and a secondary tier of replicas. The system's deep archival storage mechanism uses erasure codes to ensure data durability, and the API provides full access to OceanStore functionality, enabling applications to interact with the system effectively.