Distributed Transactions: Principles and Protocols

This lecture covers the principles and protocols of distributed transactions, including shared memory, shared disk, and shared-nothing architectures. It explains the challenges of distributed concurrency control, deadlock detection, and distributed locking. The instructor discusses the coordination of multi-operation and distributed transactions, exploring centralized and decentralized approaches. The lecture also delves into the details of distributed commit protocols like Two-Phase Commit (2PC) and the implications of coordinator failures. Additionally, it addresses the issues of restart after failures at sites, link and remote site failures, and the use of acknowledgments in the commit process. The concept of Presumed Abort and Presumed Commit in 2PC is also explained, along with the impact of coordinator failures on blocking transactions.