INTEGRATING RELIABLE MEMORY IN DATABASES

Recent results in the Rio project at the University of Michigan show t hat it is possible to create an area of main memory that is as safe as disk from operating system crashes. This paper explores how to integr ate the reliable memory provided by the Rio file cache into a database system, Prior studies have analyzed the performance benefits of relia ble memory, we focus instead on how different designs affect reliabili ty. We propose three designs for integrating reliable memory into data bases: non-persistent database buffer cache,persistent database buffer cache, and persistent database buffer cache with protection. Non-pers istent buffer caches use an I/O interface to reliable memory and requi re the fewest modifications to existing databases. However, they waste memory capacity and bandwidth due to double buffering. Persistent buf fer caches use a memory interface to reliable memory by mapping it int o the database address space. This places reliable memory under comple te database control and eliminates double buffering, but it may expose the buffer cache to database errors. Our third design reduces this ex posure by write protecting the buffer pages. Extensive fault tests sho w that mapping reliable memory into the database address space does no t significantly hurt reliability. This is because wild stores rarely t ouch dirty, committed pages written by previous transactions. As a res ult, we believe that databases should use a memory interface to reliab le memory.