Research Of Performance And Reliability Guarantee Technology Based On Buffer In ORM Middle Tier

Object Relational Mapping (ORM) is a technology used for inter-converting between object model and relation model. It can provide upper application in the form of object, which is originally represented in the form of relational data. It can also realize object persistency, namely object storage and reconstruction, in programming. The process of programming is simplified and maintainability, scalability and reusability of system are also improved while ORM is used. According to application status and requirements, research ideas that using cache in ORM to improve system performance and reliability is proposed in this thesis.Various cache technologies are comprehensively compared. Semantic cache technology is adopted for its several characters, such as flexible storage granularity, full use of storage space, and self-deduction. The SC (semantic comparability) cache replacement policy for semantic cache is proposed to take the characteristic of query locality into account. In this thesis, semantic cache is formalized and the metrics of semantic similarity are defined. This policy analyses the relativity between query items and caching items in several aspects, such as relation, condition and query attribute, and then collaborate with the data access frequency metric to determine the least similar item in cache. It can achieve higher hit ratio and lesser query response time when the queries are correlated and advent successively.In order to guarantee system reliability on the basis of improving system performance, the thesis develops a granularity adaptive consistency maintenance policy initiated by middle layer. According to the characters of semantic cache, the policy reduces the number of cache items and redundancy through combining semantic cache items. Data are processed in two ways in term of its updating frequency. Infrequently updating data uses TTL strategy, and frequently updating data uses on-demand request to keep consistency maintenance between cache and database. The granularity of updating data is changed with demand. The policy can efficiently reduce network overhead and load. Data efficiency can also be guaranteed.