动态存储的活性
On Liveness of Dynamic Storage
摘要：
动态分布式存储算法，如DynaStore，可重构Paxos，RAMBO和RDS，不确保在异步运行无限多次重新配置的活动（等待自由）。我们证明，这是固有的异步动态存储算法，包括使用$Ω$或$\diamond S$神谕。即使只有一个进程可能失败，我们的结果仍然成立，前提是成功从系统配置中删除的机器可以由管理员关闭。直觉上，不可能性依赖于这样一个事实，即正确的过程可以在任何时候被怀疑失败，即，它的故障与其他进程的消息传递缓慢是无法区分的，因此系统应该能够重新配置，而无需等待该进程完成其未决操作。为了规避这一结果，我们定义了一个动态的最终完美的故障检测器，并提出了一个算法，使用它来模拟无等待的动态原子存储（没有限制的重新配置率）。总之，我们的研究结果在$Ω$和$\diamond S$这样的预言家之间划出了一条清晰的界限，它们允许一些正确的过程永远被怀疑，而一个动态的最终完美的过程则不会。
英文摘要：
Dynamic distributed storage algorithms such as DynaStore, Reconfigurable Paxos, RAMBO, and RDS, do not ensure liveness (wait-freedom) in asynchronous runs with infinitely many reconfigurations. We prove that this is inherent for asynchronous dynamic storage algorithms, including ones that use $Ω$ or $\diamond S$ oracles. Our result holds even if only one process may fail, provided that machines that were successfully removed from the system's configuration may be switched off by an administrator. Intuitively, the impossibility relies on the fact that a correct process can be suspected to have failed at any time, i.e., its failure is indistinguishable to other processes from slow delivery of its messages, and so the system should be able to reconfigure without waiting for this process to complete its pending operations. To circumvent this result, we define a dynamic eventually perfect failure detector, and present an algorithm that uses it to emulate wait-free dynamic atomic storage (with no restrictions on reconfiguration rate). Together, our results thus draw a sharp line between oracles like $Ω$ and $\diamond S$, which allow some correct process to continue to be suspected forever, and a dynamic eventually perfect one, which does not.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
