Program guide > Programming with system APIs and plug-ins
Plug-ins for evicting cache objects
WebSphere eXtreme Scale provides a default mechanism for evicting cache entries and a plug-in for creating custom evictors. An evictor controls the membership of entries in each BackingMap. The default evictor uses a time to live (TTL) eviction policy for each BackingMap. If you provide a pluggable evictor mechanism, it typically uses an eviction policy that is based on the number of entries instead of on time.
A default TTL evictor is created with every backing map. The default evictor removes entries based on a time to live concept. This behavior is defined by the ttlType attribute, which has the following types.
Specifies that entries never expire and therefore are never removed from the map.
- Creation time
Specifies that entries are evicted depending on when they were created.
If you are using the CREATION_TIME ttlType, the evictor evicts an entry when its time from creation equals its TimeToLive attribute value, which is set in milliseconds in the application configuration. If you set the TimeToLive attribute value to 10 seconds, the entry is automatically evicted ten seconds after it was inserted.
It is important to take caution when setting this value for the CREATION_TIME ttlType. This evictor is best used when reasonably high amounts of additions to the cache exist that are only used for a set amount of time. With this strategy, anything that is created is removed after the set amount of time.
The CREATION_TIME ttlType is useful in scenarios such as refreshing stock quotes every 20 minutes or less. Suppose a Web application obtains stock quotes, and getting the most recent quotes is not critical. In this case, the stock quotes are cached in an ObjectGrid for 20 minutes. After 20 minutes, the ObjectGrid map entries expire and are evicted. Every twenty minutes or so, the ObjectGrid map uses the Loader plug-in to refresh the map data with fresh data from the database. The database is updated every 20 minutes with the most recent stock quotes.
- Last access time
Specifies that entries are evicted depending upon when they were last accessed, whether they were read or updated.
- Last update time
Specifies that entries are evicted depending upon when they were last updated.
If you are using the LAST_ACCESS_TIME or LAST_UPDATE_TIME ttlType attribute, set the TimeToLive to a lower number than if you are using the CREATION_TIME ttlType, because the entries TimeToLive attribute is reset every time it is accessed. In other words, if the TimeToLive attribute is equal to 15 and an entry has existed for 14 seconds but then gets accessed, it does not expire again for another 15 seconds. If you set the TimeToLive to a relatively high number, many entries might never be evicted. However, if you set the value to something like 15 seconds, entries might be removed when they are not often accessed.
The LAST_ACCESS_TIME or LAST_UPDATE_TIME ttlType is useful in scenarios such as holding session data from a client, using an ObjectGrid map. Session data must be destroyed if the client does not use the session data for some period of time. For example, the session data times out after 30 minutes of no activity by the client. In this case, using a TTL type of LAST_ACCESS_TIME or LAST_UPDATE_TIME with the TimeToLive attribute set to 30 minutes is appropriate for this application.
The following example creates a backing map, set its default evictor ttlType attribute, and sets its TimeToLive property.
ObjectGrid objGrid = new ObjectGrid; BackingMap bMap = objGrid.defineMap("SomeMap"); bMap.setTtlEvictorType(TTLType.LAST_ACCESS_TIME); bMap.setTimeToLive(1800);
Most evictor settings should be set before you initialize the ObjectGrid.
You may also write your own evictors: For more information, see Write a custom evictor .
The default TTL evictor uses an eviction policy that is based on time, and the number of entries in the BackingMap has no affect on the expiration time of an entry. Use an optional pluggable evictor to evict entries based on the number of entries that exist instead of based on time.
The following optional pluggable evictors provide some commonly used algorithms for deciding which entries to evict when a BackingMap grows beyond some size limit:
- The LRUEvictor evictor uses a least recently used (LRU) algorithm to decide which entries to evict when the BackingMap exceeds a maximum number of entries.
- The LFUEvictor evictor uses a least frequently used (LFU) algorithm to decide which entries to evict when the BackingMap exceeds a maximum number of entries.
The BackingMap informs an evictor as entries are created, modified, or removed in a transaction. The BackingMap keeps track of these entries and chooses when to evict one or more entries from the BackingMap.
A BackingMap has no configuration information for a maximum size. Instead, evictor properties are set to control the evictor behavior. Both the LRUEvictor and the LFUEvictor have a maximum size property that is used to cause the evictor to begin to evict entries after the maximum size is exceeded. Like the TTL evictor, the LRU and LFU evictors might not immediately evict an entry when the maximum number of entries is reached to minimize impact on performance.
If the LRU or LFU eviction algorithm is not adequate for a particular application, you can write the own evictors to create your eviction strategy.
Memory-based eviction is only supported on Java™ Platform, Enterprise Edition v5 or later.
All built-in evictors support memory-based eviction that can be enabled on the BackingMap interface by setting the evictionTriggers attribute of BackingMap to MEMORY_USAGE_THRESHOLD. For more information about how to set the evictionTriggers attribute on BackingMap, see BackMap interface and ObjectGrid descriptor XML file.
Memory-based eviction is based on heap usage threshold. When memory-based eviction is enabled on BackingMap and the BackingMap has any built-in evictor, the usage threshold is set to a default percentage of total memory if the threshold has not been previously set.
When you are using memory-based eviction, you should configure the garbage collection threshold to the same value as their target heap utilization. For example, if the memory-based eviction threshold is set at 50 percent and the garbage collection threshold is at the default 70 percent level, then the heap utilization can go as high as 70 percent. This heap utilization increase occurs because memory-based eviction is only triggered after a garbage collection cycle.
The memory-based eviction algorithm used by WebSphere eXtreme Scale is sensitive to the behavior of the garbage collection algorithm in use. The best algorithm for memory-based eviction is the IBM default throughput collector. Generation garbage collection algorithms can cause undesired behavior, and so you should not use these algorithms with memory-based eviction.
To change the usage threshold percentage, set the memoryThresholdPercentage property on the container and server property files for eXtreme Scale server processes.
During runtime, if the memory usage exceeds the target usage threshold, memory-based evictors start evicting entries and try to keep memory usage below the target usage threshold. However, no guarantee exists that the eviction speed is fast enough to avoid a potential out of memory error if the system runtime continues to quickly consume memory.
- TimeToLive (TTL) evictor
WebSphere eXtreme Scale provides a default mechanism for evicting cache entries and a plug-in for creating custom evictors. An evictor controls the membership of entries in each BackingMap instance.
- Plug in a pluggable evictor
Since evictors are associated with BackingMaps, use the BackingMap interface to specify the pluggable evictor.
- Write a custom evictor
WebSphere eXtreme Scale allows you to write a custom eviction implementation.
- Plug-in evictor performance best practices
If you use plug-in evictors, they are not active until you create them and associate them with a backing map. The following best practices will increase performance for least frequently used (LFU) and least recently used (LRU) evictors.
Parent topic:Program with system APIs and plug-ins
Introduction to plug-ins
Plug-ins for transforming cached objects
Plug-ins for versioning and comparing cache objects
Plug-ins for custom indexing of cache objects
Plug-ins for communicating with persistent stores
Plug-ins for managing transaction life cycle events
Plug-ins for providing event listeners