10.07.2015 Views

Expert Oracle Exadata - Parent Directory

Expert Oracle Exadata - Parent Directory

Expert Oracle Exadata - Parent Directory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CHAPTER 7 RESOURCE MANAGEMENT-- IORM Database View (interdatabase IORM metrics)CREATE OR REPLACE FORCE VIEW RJOHNSON.DB_IORM_METRICS ASSELECT name,OBJECT_TYPE,METRIC_NAME,SUM (TO_NUMBER (TRIM (METRIC_VALUE))) metric_value,MAX (INC) incFROM iorm_metricsWHERE name LIKE 'DB\_%' ESCAPE '\' IORM Database recordsAND name NOT LIKE 'DB\_FC\_%' ESCAPE '\'AND name NOT LIKE 'DB\_FD\_%' ESCAPE '\'AND METRIC_NAME IN ('SCRATCH', 'SNIFF')GROUP BY name, OBJECT_TYPE, METRIC_NAMEORDER BY name, OBJECT_TYPE, METRIC_NAME;-- IORM Consumer Group View (intradatabase IORM metrics)CREATE OR REPLACE FORCE VIEW RJOHNSON.CG_IORM_METRICS ASSELECT name,OBJECT_TYPE,METRIC_NAME,SUM (METRIC_VALUE) metric_value,MAX (INC) incFROM iorm_metricsWHERE name LIKE 'CG\_%' ESCAPE '\' IORM Consumer Group recordsAND name NOT LIKE 'CG\_FC\_%' ESCAPE '\'AND name NOT LIKE 'CG\_FD\_%' ESCAPE '\'AND (metric_name LIKE 'SCRATCH%' OR metric_name LIKE 'SNIFF%')AND metric_name NOT LIKE '%\_' ESCAPE '\'GROUP BY name, OBJECT_TYPE, METRIC_NAMEORDER BY name, OBJECT_TYPE, METRIC_NAME;You may have noticed the SUM(METRIC_VALUE) function in the view definitions. While it is important tounderstand the workload mix of your databases, the purpose of this exercise is to verify the effectivenessof our IORM Resource Plan. To that end, we’ll subtotal the small and large I/O metrics so that we can seehow all I/O is distributed. Importing the IORM metrics from our table into a spreadsheet, we can createpie graphs for each IORM consumer group. The resulting graphs should reflect the proportions for eachIORM consumer in our IORM plan. There are several interesting metrics captured in the METRICCURRENTobject at the storage cell, but we’ll focus on IO_BY_SEC. This metric represents the I/O requests that weresent to the cell disk queues for execution. It should give us a pretty good idea of how I/O requests wereprioritized.One more thing before we take a look at the results of our test. While the filters we applied to ourviews simplified the output, they also eliminated some I/O from the results. For example, the_ORACLE_BACKGROUND_GROUP_ metric accounted for a measurable amount of I/O in the storage cells but isnot reflected in the analysis. That’s acceptable, though, because the point of this exercise is not toprecisely reconcile the results of our test with the IORM allocations called for in our plan directives (avery tedious process). Rather, the purpose is to show how I/O is actually allocated in the storage cellsduring peak periods when IORM is active. So while the results are expected to resemble our IORM plans,they are not expected to match precisely.229

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!