Chapter 32. Set Up the L1 Cache
32.1. About the L1 Cache
The Level 1 (or
L1) cache stores remote cache entries after they are initially accessed, preventing unnecessary remote fetch operations for each subsequent use of the same entries. The L1 cache is only available when Red Hat JBoss Data Grid’s cache mode is set to distribution. In other cache modes any configuration related to the L1 cache is ignored.
When caches are configured with distributed mode, the entries are evenly distributed between all clustered caches. Each entry is copied to a desired number of owners, which can be less than the total number of caches. As a result, the system’s scalability is improved but also means that some entries are not available on all nodes and must be fetched from their owner node. In this situation, configure the Cache component to use the L1 Cache to temporarily store entries that it does not own to prevent repeated fetching for subsequent uses.
Each time a key is updated an invalidation message is generated. This message is multicast to each node that contains data that corresponds to current L1 cache entries. The invalidation message ensures that each of these nodes marks the relevant entry as invalidated. Also, when the location of an entry changes in the cluster, the corresponding L1 cache entry is invalidated to prevent outdated cache entries.
32.2. L1 Cache Configuration
32.2.1. L1 Cache Configuration (Library Mode)
The following sample configuration shows the L1 cache default values in Red Hat JBoss Data Grid’s Library Mode.
L1 Cache Configuration in Library Mode
<distributed-cache name="distributed_cache" l1-lifespan="0" l1-cleanup-interval="60000"/>
The following attributes control the L1 cache behavior:
l1-lifespanattribute indicates the maximum lifespan in milliseconds of entries placed in the L1 cache, and is not allowed in non-distributed caches. By default L1 this value is 0, indicating that L1 caching is disabled, and is only enabled if a positive value is defined.
li-cleanup-intervalparameter controls how often a cleanup task to prune L1 tracking data is run, in milliseconds, and by default is defined to 10 minutes.
32.2.2. L1 Cache Configuration (Remote Client-Server Mode)
The following sample configuration shows the L1 cache default value of
0, indicating it is disabled, in Red Hat JBoss Data Grid’s Remote Client-Server mode:
L1 Cache Configuration for Remote Client-Server Mode
<distributed-cache l1-lifespan="0"> <!-- Additional configuration information here --> </distributed-cache>
l1-lifespan element is added to a
distributed-cache element to enable L1 caching and to set the life span of the L1 cache entries for the cache. This element is only valid for distributed caches.
l1-lifespan is set to
0 or a negative number (
-1), L1 caching is disabled. L1 caching is enabled when the
l1-lifespan value is greater than
When the cache is accessed remotely via the Hot Rod protocol, the client accesses the owner node directly. Therefore, using L1 Cache via the Hot Rod protocol is not recommended; instead, refer to the Near Caching section in the JBoss Data Grid Developer Guide . Other remote clients (Memcached, REST) cannot target the owner, therefore, using L1 Cache may increase the performance (at the cost of higher memory consumption).
In Remote Client-Server mode, the L1 cache was enabled by default when distributed cache was used, even if the
l1-lifespan attribute is not set. The default lifespan value was 10 minutes. Since JBoss Data Grid 6.3, the default lifespan is 0 which disables the L1 cache. Set a non-zero value for the
l1-lifespan parameter to enable the L1 cache.