Cluster-Wide Analysis of the Cache Load

Interval Between Two Snapshots in Minutes

Interval between two snapshots specified in minutes.

While the report is running, snapshots are taken of the cache use in the defined interval. The measurements in all snapshots are then analyzed in the results.

Minimum interval: 5 minutes

Maximum interval: 60 minutes

Number of Snapshots

Number of snapshots.

Once all snapshots have been taken, the report calculates the results and recommendations.

Condense Caches of CM Repositories

In the configuration of CM repositories, there is a default cache that is used as a template and from which you can create your own caches at runtime with a predefined naming pattern. (You can change only the template in the configuration.)

When calculating the optimization proposals, the report can take into account the caches created at runtime individually. You then receive proposals that you cannot implement because these caches do not exist in the KM configuration.

The benefit of this setting is that you receive more information about the load on the individual caches if more than one CM repository uses the same template.

Depending on the results, you can decide whether or not to stop using a shared cache and instead create separate caches for some CM repositories to spread the cache load more efficiently.

If you activate this parameter, you receive proposals for the templates. The proposals are based on the evaluation off all CM repositories that use this template.

Ignore Snapshots Taken in Idle Periods

If this parameter is activated, snapshots taken during idle periods, for example, during scheduled downtimes, are not taken into account.

A period is defined as idle time if the number of gets in the snapshot for that period contains less than ten percent of the average number of gets for all snapshots.

Item

This column does not contain any information that can be analyzed.

Description

This column does not contain any information that can be analyzed.

Name, Capacity, Current, Peak, Gets, Hits, Hits/Gets

See Cache Monitor .

Proposal

See the section below.

+

The cache is used intensively, therefore you should increase the capacity of the cache. In most cases, this should have a positive impact on the system performance. Note that an increase in cache capacity is linked to an increased requirement for memory resources.

Increasing the capacity of a cache that is used completely and intensively speeds up a system. However, this measure also takes up additional memory in which the new entries ultimately have to be stored. Therefore we recommend increasing the capacity incrementally, for example, in 20% steps. After each increase, check whether or not the system is still stable or has reached its memory limit.

Recommendations for increasing the capacity are based on the following measurements:

• The number of cache entries has filled the capacity of the cache at least once.
• The ratio of hits to gets is lower than 90%.
• The capacity of the cache is smaller than 15000.
• The average number of entries corresponds to at least 90% of the capacity or reached at least 90% of the capacity in 30% of all snapshots.

+/- 0

The cache load is sufficient. No changes are necessary.

-( → followed by a value)

There is only a small load on the cache, therefore the capacity of the cache can be reduced by this value.

Note that the reduction of the capacity of a cache that contains few entries hardly saves memory. It only limits the consumption in extreme cases.

Recommendations for reducing the capacity are based on the following measurements:

• The number of cache entries has never filled the capacity of the cache.
• The average number of cache entries is below 90% of the capacity.
• The capacity of the cache is larger than 100.

The recommended value is an average that is calculated from the average number of entries and the maximum number of entries in the cache.