Assuming you don’t know anything literally about the Oracle Real Application Clusters (RAC) environment, You should start with:
1. Database version, O/S version, how many nodes, tns names, listener information?
2. Database Size, Storage architecture, raid/ASM info. ?
3. Details on VIPs, physical IPS, server names, kernel information, db parameters,
4. Backup and restore strategy
5. Application information which database supports and architecture information of application and connectivity details. Any special requirements from application regards to connectivity or configuration?
6. A good architecture diagram should provide most of the above details. Details on your database/application users,
7. List of standard or customized jobs that run the database daily, weekly and monthly.
8. Current change management procedure/controls and their schedules.
9. Database growth rate/month?
I also recommend you obtain Automatic Workload Repository (AWR) and Automatic Database Diagnostic Monitor (ADDM) reports. The AWR and ADDM reports should be representative of a time of heavy or peak load. You want to know what is causing stress on the system, information that during off peak time gives you very little to work on. Some areas I recommend that you key on when reviewing the AWR include:
a. Number of instances
b. Instance global cache load profile
c. Global cache efficiency percentage
d. GCS and GES – workload characteristics
e. Messaging statistics
f. Service statistics
g. Service wait class statistics
h. Top 5 CR and current block segments
Always look at the TOP 5 Events! This will give you a very clear understanding of issues during the peak
load time. Some things you might see include:
a. DB CPU
b. db file scattered read
c. direct path read
d. db file sequential read
e. log file sync
Always check your Interconnect.
c. Receive times
d. 1gb interconnect = 100 MB/sec (approx)
e. If interconnect latency>IO subsystem latency, RAC is a bottleneck
I also recommend that you obtain an ADDM report from the AWR that you have. This report will give you additional information that will help you understand what is going on with the system. ADDM ranks both the problems and its recommendations according to the crucial DB time statistic. Every time AWR runs ADDM automatically does a top-down system analysis and reports its findings on the database control home.
ADDM will report on the following:
1. Expensive SQL/Java statements
2. I/O performance issues
3. Locking and Concurrency issues
4. Excessive parsing
5. High checkpoint load
6. Resource bottlenecks, including memory and CPU bottlenecks
7. Undersized memory allocations
8. Connection management issues, such as excessive logon/logoff activity
The report itself will contain
1. Expert problem diagnosis
2. Emphasis on the root cause of the problem rather than on the symptoms
3. A ranking of the effects of the problems, which means you can quickly find the problem
4. Recommendations ranked according to their benefit