Owners Manual

Optimizing Your System | Getting Started
114 OMNM 6.5.2 User Guide
BUFFER POOL AND MEMORY
----------------------
Buffer pool hit rate 1000/1000
You may need to modify system settings, increase or decrease application server heap,
web server heap, and innodb buffer to fit your needs. This depends on whether you use
the webserver heavily.
2. innodb_log_file_size = 256 M to 1024m
A larger file improves performance, but setting it too large will increase recovery time in
case of a crash or power failure. Best practice is to experiment with various settings to
determine what size is best for performance.
To change the log file size, you must move existing the log files named ib_logfile0,
ib_logfile1, and so on. See Changing InnoDB Log Files in MySQL on page 116 for step-
by-step instructions. The database may not start if you configure a log file size
mismatch.
3. max_connections=100 to 1000
Best practice is to configure 200 or more connections per server (application server +
web server), especially if you are adding more servers.
The number of connections permitted in this version of MySQL defaults to 100. If you
need to support more connections, set a larger value for this variable. Windows is
limited to (open tables × 2 + open connections < 2048) because of the Posix
compatibility layer used on that platform.
Log in to mySQL to check current settings:
mysql -u root --password=dorado
mysql> show variables like 'max_connections';
To check open connections:
mysql> SHOW STATUS WHERE `variable_name` = 'Threads_connected';
NOTE:
You may need to reduce table_cache if you increase max_connection.
4. table_cache = 1024
(increase the default as appropriate).
table_cache is related to max_connections. For example, for 200 concurrent
running connections, you should have a table cache size of at least 200 * N, where N is
the maximum number of tables per join in any of the queries which you execute. You
must also reserve some extra file descriptors for temporary tables and files.
If the value is very large or increases rapidly, even when you have not issued many
FLUSH TABLES statements, you should increase the table cache size.
5. Monitors: If you enabled and configured the default SNMP interface monitor, it would
typically consume most of the space in owbusdb (the database).
Unless you have reason to do otherwise, best practice is to disable Retain polled data on
the default SNMP interface monitors. The graphs do not need these data for display.
OpenManage Network Manager only uses retained data to derive the calculated
metrics attributes. In most cases, saving only calculated data for the default SNMP
interface monitor suffices.