As the significantly acceleration of database and cloud computing technology, data centers have become increasingly integral to vital businesses in various industries, including government departments and enterprise institutions. Data quality issues are magnified in such circumstances. However, due to an extensive amount of data access and the intricate cross- domain environment, there are inevitable data loss problems in the access process. From the perspective of improving the observability of the whole data flow, this paper proposes a log- based data access loss detection (LDLD) model and constructs a prototype system. The experimental results in the actual system show that the LDLD model can effectively detect packet loss during data access and accurately locate the fault location. This study can provide key technical support for solving the data quality problems of data centers.