Monitoring Exadata Storage Cells - Oracle Exadata Recipes: A Problem-Solution Approach - page 377

N_HCA_MB_RCV_SEC cm01cel01 0.156 MB/sec Rate
N_HCA_MB_TRANS_SEC cm01cel01 0.151 MB/sec Rate
... Additional metrics omitted
CellCLI>
The metrichistory Object
The metrichistory object contains, by default, seven days of historical metrics sampled over one-minute intervals.
To display the attributes of the metrichistory object, use the following describe command in CellCLI:
CellCLI> describe metrichistory
name
alertState
collectionTime
metricObjectName
metricType
metricValue
metricValueAvg
metricValueMax
metricValueMin
objectType
CellCLI>
You will notice that the metrichistory object contains the same attributes as the metriccurrent object, plus
average, minimum, and maximum values of the metricValue attribute. Using the same CL_CPUT metric demonstrated
previously, run the following CellCLI command to display your historical storage server CPU utilization:
CellCLI> list metrichistory CL_CPUT
CL_CPUT cm01cel01 3.6 % 2012-10-18T00:00:59-04:00
CL_CPUT cm01cel01 0.4 % 2012-10-18T00:01:59-04:00
CL_CPUT cm01cel01 1.7 % 2012-10-18T00:02:59-04:00
... Additional output omitted for brevity
As mentioned, Exadata's storage server historical metrics store information at one-minute intervals and the
volume of information that an Exadata DMA would need to analyze can often be challenging. You can add where
conditions on the collectionTime attribute to limit the output to specific date ranges. Using our CL_CPUT metric,
execute the following CellCLI command to introduce a date-based search criteria:
CellCLI> list metrichistory CL_CPUT where collectionTime < '2012-10-18T00:07:59-04:00';
CL_CPUT cm01cel01 3.6 % 2012-10-18T00:00:59-04:00
CL_CPUT cm01cel01 0.4 % 2012-10-18T00:01:59-04:00
CL_CPUT cm01cel01 1.7 % 2012-10-18T00:02:59-04:00
CellCLI>
Experienced Exadata DMAs typically employ a number of techniques to make the volume of metrichistory data
easier to cope with. Recipe 13-6 includes some techniques for summarizing large amounts of metrichistory data.
The alertState Attribute
One of the more important metric object attributes, from a monitoring standpoint, is the alertState attribute.
Exadata DMAs often will build reporting or monitoring strategies that include reporting on some or all metrics with
an abnormal alertState . You can find any metrics in an abnormal state from current metrics (that is, metriccurrent )