oracle 成长历程-3.3 影响buffer cache的内部参数

3.3     影响buffer cache的内部参数

n   _DB_BLOCK_HASH_BUCKETS: block hash buckets的数量 (默认是DB_CACHE_SIZE/4)

n   _DB_BLOCK_MAX_SCAN_CNT: 寻找空闲buffer的最大尝试次数

n   _DB_WRITER_SCAN_DEPTH: DBWR寻找脏块的的扫描深度

n   _DB_BLOCK_CACHE_CLONE: 取的时候，复制一个副本 (调试用)

n   _DB_BLOCK_CACHE_MAP: Map/unmap and track reference counts on blocks (for debugging purposes only)

n   _DB_BLOCK_MAX_CR_DBA: Maximum allowed number of CR buffers per dba

n   _DB_WRITER_HISTOGRAM_STATISTICS: Maintain DBWR histogram statistics in X$KCBBHS

n   _DB_BLOCK_MED_PRIORITY_BATCH_SIZE: Percentage of write batch used for medium priority checkpoints

n   _DB_BLOCK_HI_PRIORITY_BATCH_SIZE: Percentage of write batch used for high priority checkpoints

n   _DB_PERCENT_HOT_DEFAULT: Percent of default buffer pool considered hot

n   _DB_PERCENT_HOT_KEEP: Percent of keep buffer pool considered hot

n   _DB_PERCENT_HOT_RECYCLE: Percent of recycle buffer pool considered hot

n   _DB_AGING_HOT_CRITERIA: Touch count which sends a buffer to head of replacement list

n   _DB_AGING_COOL_COUNT: Touch count set when buffer cooled

n   _DB_AGING_TOUCH_TIME: Touch count which sends a buffer to head of replacement list

n   _DB_AGING_FREEZE_CR: Make CR buffers always be too cold to keep in cache

n   _DB_AGING_STAY_COUNT: 当buffer移动到LURheader的Touch count

n   _DB_BLOCK_CACHE_PROTECT: 保护 database blocks (调试用)

n   _DB_BLOCK_HASH_LATCHES: database block hash latches的数目

n   _DB_HANDLES: 同时能发生的buffer操作，系统级

n   _DB_HANDLES_CACHED: 每个进程缓存的Buffer 句柄

n   _DB_LARGE_DIRTY_QUEUE: 脏块队列强制写入磁盘的门限

n   _DB_WRITER_MAX_WRITES: 最大的DB Writer I/Os

n   _DB_WRITER_CHUNK_WRITES: DBWR 尝试写等待次数

n   _DBWR_ASYNC_IO: 激活 DBWR 异步写

n   _DBWR_TRACING: 激活 DBWR 追踪

n   _DBWR_SCAN_INTERVAL: DBWR scan interval

n   _TRACE_BUFFER_FLUSHES: 追踪buffer flushes 如果otrace cacheIO事件被设置

n   _TRACE_MULTI_BLOCK_READS: 追踪 multi-block 读，如果otrace cacheIO事件被设置

n   _TRACE_CR_BUFFER_CREATES: 追踪 CR buffer 创建，如果 otrace cacheIO 事件被设置

n   其它技巧

n   使用多个LRU latch的时候，使用多个DBWR进程。

n   尽量使LRU latch丢失率小于1％

n   使用asynchronous I/O 或者 DBWR I/O Slaves.

n   减少 buffer cache load 通过使用direct I/O.

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/347643/viewspace-619861/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/347643/viewspace-619861/