内存受限的实时内存数据库数据装入策略

A Data Loading Scheme for RTMMDB with Limited Main Memory

实时数据库要求以内存数据库为其底层的支持。传统的内存数据库要求内存数据库能容纳全部外存数据库,因此没有数据装入的问题。而要实现内存受限的内存数据库,就是在内存数据库不能容纳全部外存数据库的前提下,保证一个事务执行前,其数据被装入内存。而内存数据库的脆弱性使得数据的装入比较频繁(包括初装和运行时装入),因此传统的装入策略对实时内存数据库就显得不合适。为此给出了新的数据装入策略,并在此基础上实现了ERTMMDB(Embedded Real-time Main Memory Database )。

A real-time database requires a main memory database(MMDB) to support it as its base. The traditional MMDB assumes that the entire database is stored in volatile main memory, so there is no trouble of data loading. But the MMDB with limited main memory should assure that before running a transaction, all data accessed by it should be loaded into the main memory. The vulnerability of a MMDB makes its loading (including initial loading and running load) very frequent. Traditional database loading approaches are not suitable for RTMMDB. In this paper, a data loading scheme for RTMMDB with limited main memory is presented, based on it, an ERTMMDB (embedded real-time main memory database) is developed.