Spotlight on SAP ASE 2.12 - User Guide

CPU Busy Alarm

This alarm is raised when the percentage of time that the Adaptive Server's CPU was performing Adaptive Server-related tasks has crossed a defined threshold.

By checking Spotlight for problems, and tuning to alleviate contention, response time can remain high even at "CPU Busy" values in the 80 to 90% range.

Note: When engine utilization is extremely high for an extended time, the housekeeper process writes few or no pages out of disk (since it runs only during idle CPU cycles). This means that when a checkpoint does get to run, it may find many pages that need to be written to disk (i.e., after a large batch job or a database dump). The checkpoint process is likely to send CPU usage to 100% for a period of time, causing a perceptible dip in response time.

If values are consistently very high (more than 90%), it is likely that response time and throughput could benefit from an additional engine. If you have an SMP box and the available CPUs, and if you elect to improve response time and throughput by adding Adaptive Server engines, check for an increased resource contention in other areas after adding each engine.

Data Cache Hit Rate Alarm

This alarm is raised when the data cache hit rate for the server falls below the ranges defined by the thresholds on this component.

This hit rate is calculated as a percentage of searches that resulted in the needed page being found in the data cache (a cache hit) from the total number of searches to the data cache.

Use the Memory drilldown | Data Cache page to find out individual cache hit rates and to find any specific cache that may be bringing the total server percentage down.

Data Cache Reads Alarm

This alarm is raised when the rate of physical reads requested by the data cache (expressed as reads per second) exceeds normal levels.

Since this metric is included in calibration, the maximum value, and thereby the alarm thresholds and flow speed, are determined by statistical sampling.

This rate is determined by cache searches that did not find the needed page in the data cache (a cache miss).

Use this data, along with Cache Searches, to determine how effective the overall cache design for this server is.

Use the Memory drilldown | Data Cache page to find out which cache has the most utilization. After investigation, you may decide that a separate named cache should be created to alleviate any detected contention.

Data Cache Searches Alarm

This alarm is raised when the rate of searches requested from the data cache (expressed as searches per second) exceeds normal levels.

Since this metric is included in calibration, the maximum value, and thereby the alarm thresholds and flow speed, are determined by statistical sampling.

This rate includes searches that resulted in the needed page being found in the data cache (a cache hit) and those searches that did not find the page in cache and required a disk read (a cache miss).

Cache searches are one indication of server activity.

Use this data together with Physical Reads to determine how effective the overall cache design for this server is.

Use the Memory drilldown | Data Cache page to find out which cache has the most utilization. After investigation, you may decide that a separate named cache should be created to alleviate any detected contention.