![]() ![]() Inspired by the M&M algorithm, we design and implement LCL, an elegant and generally applicable algorithm for resource deadlock detection and resolution in distributed environments without a restriction of the above kind. Mitchell and Michael J Merritt (M&M) proposed a simple and fully distributed deadlock detection and resolution algorithm, but the assumption that each process waits for only one resource each time prevents the algorithm from being used in many usage scenarios. ![]() While it has long been a mature feature of classical centralized database systems for many years, its use in distributed database systems remains in its infancy. The problem of deadlock detection and resolution in database systems has been studied for decades. Extensive emulation experiments and practices in OceanBase Database prove that the LCL algorithm significantly outperforms the M&M algorithm in the efficiency of deadlock detection and resolution. Inspired by the M&M algorithm, this paper presents the elegant and generally applicable LCL algorithm for resource deadlock detection and resolution in distributed environments. As a result, this algorithm cannot be used in many usage scenarios. Merritt (M&M) proposed a simple and fully distributed deadlock detection and resolution algorithm based on the assumption that each vertex waits for only one resource each time. The distributed environment brings a new challenge for global deadlock detection and resolution. Abstractĭeadlock detection and resolution have been important research subject for classical relational databases for years. This paper provides a rigorous mathematical proof of the correctness of LCL and describes experiments based on a distributed transaction processing emulator (TPE) to prove the efficiency and scalability of LCL. LCL can detect and resolve all real deadlocks in the system without inducing redundant interruptions. ![]() This consumes a small amount of memory and communications resources. In LCL-based deadlock detection, a transaction process (vertex) does not need a global wait for graph (WFG) view or a local view, but only needs to send dozens of bytes of data to the vertices that hold the resources for which the current vertex is waiting. In most cases, LCL is more suitable than other deadlock detection algorithms and can be easily configured in all types of relational database management systems (RDBMSs). OceanBase Database proposed a lock chain length (LCL)-based deadlock detection and resolution algorithm for distributed database systems. The acceptance of this paper marks that OceanBase Database provides a leading solution to distributed deadlock detection. The papers that are submitted to the ICDE research track must be original and describe deep theoretical insights. Compared with the industry track, the research track module is designed to track the achievements of academic and industrial researchers and focuses more on the latest progress in data engineering. OceanBase Database and Alibaba Cloud are the only companies that contributed papers to the research track of ICDE this year. Most papers are contributions from famous universities and scientific research institutions worldwide. ICDE, Special Interest Group on Management of Data (SIGMOD), and the Very Large Database (VLDB) conference are the top conferences for database-related fields.Ī total of 228 papers are accepted by the research track of ICDE 2023. IEEE ICDE is an important international academic conference for databases. The algorithm can help improve the reliability and performance of a distributed database system. The innovative algorithm that is proposed in this paper can accurately detect and resolve all real deadlocks in a distributed environment. A paper titled “LCL: A Lock Chain Length-based Distributed Algorithm for Deadlock Detection and Resolution” was recognized at International Conference on Data Engineering (ICDE) 2023. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |