1.Global Cache Load Profile

Global Cache blocks received:  接收到的全局缓冲块

Global Cache blocks served: 发送的全局缓冲块

GCS/GES messages received: GCS消息接收

GCS/GES messages sent: GCS消息发送

DBWR Fusion writes: 其他节点要求写操作

Estd Interconnect traffic (KB)： 预计带宽占用

(计算公式:Estd Interconnect traffic (KB) = (('gc cr blocks received'+ 'gc current blocks received' + 'gc cr blocks served'+ 'gc current blocks served') * Block size) + (('gcs messages sent' + 'ges messages sent' + 'gcs msgs received'+ 'gcs msgs received')*200)/1024/Elapsed Time)

反应Private Network性能的2个维度：

a)可用带宽

b)网络延迟

2.Global Cache Efficiency Percentages

local cache: 本地缓冲区获取数据

remote cache: 远程缓冲区获取数据

disk: 磁盘获取数据

3.Global Cache and Enqueue Services - Workload Characteristics

Avg global enqueue get time (ms): 一个global cache获取时间1~20ms正常。

Avg global cache cr block receive time (ms): 平均每个global cr块从申请到收到的耗时。2ms正常

Avg global cache current block receive time (ms): 平均每个global current块从申请到收到的耗时。2ms正常

gc cr block flush time

Flush time是在redo块上的操作。flush 是Oracle为了保证Instance Recovery实例恢复机制，而要求每一个current block在本地节点local instance被修改后(modify/update) 必须要将该current block相关的redo 写入到logfile 后（要求LGWR必须完成写入后才能返回)，才能由LMS进程传输给其他节点使用。RAC 的Redo flush慢造成gc buffer busy release/acquire等待

CR、Current两者的flush指标一般都要求<5ms，在Global CURRENT Served Stats中可以看到 current块的flush时间分布在哪些范围内

gc cr block build time

gc cr block send time

gc current block pin time

gc current block flush time

gc current block send time

4.Global Cache and Enqueue Services - Messaging Statistics 

Avg message sent queue time (ms):非直接传输的消息，在队列中的等待时间，0.1ms左右

Avg message sent queue time on ksxp (ms):ipc层面发送一个消息到ack回应，应小于1ms

Avg message received queue time (ms):消息进入队列到开始处理的时间，0.1ms左右

Avg GCS message process time (ms):处理一个buffer相关的gcs时间，应小于0.5ms

Avg GES message process time (ms):处理一个锁消息的时间，应小于0.5ms

% of direct sent messages:直接发送消息

% of indirect sent messages:通过代理发送消息

% of flow controlled messages:流控消息

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/15412087/viewspace-2199783/，如需转载，请注明出处，否则将追究法律责任。