Tools and Utilities - Expert Oracle RAC Performance Diagnostics and Tuning - page 169

There is a subtle difference between Oracle Database 11g Release 2 and Oracle Database 12c Release 1. In the
earlier version, the repository used to store the diagnostic data was a Berkelydb compared to the Oracle Database 12c
Release 1, called the management instance installed as part of the (GI) installation.
CHM provides real-time monitoring, continuously tracking O/S resource consumption at node, processes, and
device levels to help diagnose degradation and failures of the various components in a RAC environment. In the
versions 10g and 11g, the real-time data was available through a GUI interface or historical data could be replayed
from the repository stored on individual nodes in the cluster to analyze issues such as high interconnect latency times,
high CPU usage, runaway processes, high run queue lengths, and off course reasons and causes that lead to situations
causing node evictions.
In Oracle Database 12c Version 1, only a command-line interface is available to analyze data.

Architecture
Similar to the Oracle clusterware, the CHM architecture is robust (as illustrated in Figure 6-21 ). CHM has several
daemon processes that run on all servers to perform various tasks.

Node 1 Node 2 Node 3 Node 4 Node 5 Node 6

SSKY1 SSKY2 SSKY3 SSKY4 SSKY5 SSKY6

ologgerd
osysmond

ologgerd
osysmond osysmond osysmond osysmond osysmond

AV7 AV8 AVL9 AV10 AV11 AV12 AV13 AV14 AV15 AV16

GRID_DATA PRD_DATA

Figure 6-21. CHM Oracle Database 12c architecture

CHM consists of a database and three daemons:

•

oproxyd is the proxy daemon that handles connections (prior to Oracle 12c). All connections
are made via the public interface. If CHM is configured to use a dedicated private interface,
then oproxyd listens on the public interface for external clients such as oclumon and crfgui
(discussed later in this section). It is configured to run on all nodes in the cluster.

•

osysmond is the primary daemon responsible for monitoring and O/S metric collection. It's
responsible for metrics collection on all nodes in the cluster and sending the data to the
ologgerd daemon process. osysmond is responsible for spawning/starting the ologgerd , and
for some reason if the ologgerd dies, it will restart it.