学术界和工业界提出了许多面向OLTP的图数据库和面向OLAP的图计算系统，但这两类系统目前仍然各自独立发展。图计算系统需要经过ETL过程，从OLTP系统中批量导出静态的图数据，进而在静态的图数据上执行相应的图算法。正如传统数据仓库面临的痛点一样，图计算系统无法依据最新的数据进行分析决策。为此，本项目探索面向混合OLTP和OLAP的分布式图数据库系统构建。针对混合系统的正确性问题，基于MVCC和TO并发控制机制，我们提出了一套全新的面向混合OLTP事务操作与OLAP分析任务的正确性理论。为了解决混合系统的高效性问题，我们设计了新的事务调度策略，并从冲突检测、事务回滚、事务重做三个方面对事务调度进行了优化，并从图存储模式、图划分、新硬件加速三个方面对存储进行了优化。我们拟通过以上研究，解决混合系统的正确性和高效性问题。因此，本项目是对现有图计算系统和图数据库的重要补充。

Extensive OLTP graph databases and OLAP graph computing systems have been proposed in academia and industry, while they are still developing independently. Graph computing systems need to go through ETL processes to export static graph data in batches from OLTP systems, and then execute corresponding graph algorithms on the static graph data. Just as the traditional data warehouse systems are faced with, graph computing systems cannot analyze and make decisions based on the latest data. Targeting on this issue, this project explores to build distributed OLTP and OLAP database systems. In order to guarantee the correctness of our hybrid system, we propose a series of correctness theories on the hybrid OLTP transaction operations and OLAP analysis based on MVCC and TO concurrency control mechanism. To improve efficiency of our hybrid system, we designed a new transaction scheduling strategy which is optimized in terms of conflict detection, transaction rollback and transaction replaying. We also optimized the distributed graph storage engine in terms of graph storage mode, graph partition and acceleration based on new hardware. Thus, we try to address the correctness and efficiency issues of the hybrid system via the above methodology. To conclude, our project is positioned as an important complement of the existing graph computing systems and graph databases.