Implementation Of NoSQL Database Distributed Cache System

With the arrival of the WEB2.0era, the amount of data in the Internet isexploding. The frequent access of huge amount of data has been challenging theperformance of data servers and database. And the method of further enhancingthe efficiency of storing and searching massive data, without causing much stressto the data servers, is becoming key research contents in big data fields. To set upa distributed cache system upon the data servers cluster is a preferable solution tothe problem raised above.On the international and domestic market nowadays, there are a variety ofcommercial and open-source cache products. However, most of them haveshortcomings to varying degrees on the aspect of performance and scope ofapplication, such as weak autonomy ability, low cache hit ratio or not applicable tosome application situation.Therefore, in this issue, a kind of cache system, which is of good architecture,for NoSQL database under distributed environment has been designed based onthe project that is developed by and for the actual demand of the enterprise of theinformation system construction.In the “cluster memory self-adaptive management” module of the system, theinnovative method of memory pre-allocation and self-adaptive adjustment is putforward to improve the system s memory utilization, capacity of autonomy andcache hits ratio. While in the load balancing aspect, the concept of “Bucket” is putforward to act as virtual node and a data distributing table is maintained in theConfigure Server, instead of classical consistent hashing, which increases thesystem s ability of centralized management and lies a good foundation for futuredevelopment on the module of persistent storage.In this issue, the design idea of the whole system and implementation detailsof some key components is illustrated, in addition, the test work and its resultsanalysis of the related function and performance of the system is done as well.