XAP as Hibernate Second Level Cache

This page explains how you can use XAP to:

• Get a free 2nd-level distributed cache for your Hibernate-based database integration
• Scale up your database using the existing ORM by adding a GigaSpaces Data Grid with back-end Hibernate integration

GigaSpaces as Hibernate 2nd level cache provides:

• Unlimited cache capacity
• All activities be monitored/profiled in real-time
• Built-in support for replicated and/or partitioned cache topology
• Built-in support for embedded or remote cache
• Built-in dynamic scalability
• Built-in High-Availability support
• Built-in Spring integration
• Built-in WAN replication
• Built-in security
• Built-in deployment environment
• Built-in alerting system

Read further in this section for more details about these solutions, or jump to detailed instructions: how to get a free distributed second-level cache or how to add a GigaSpaces data grid.

The Existing Architecture

Your existing Hibernate-based database integration probably looks like the following: the application uses Hibernate to persist its data directly to the database.

Step 1: Getting a Free Distributed Two-Level Cache

You can very easily replace your existing Hibernate cache provider, typically EHCache, with GigaSpaces. This gives you a powerful distributed cache with multiple clients.

Benefits:

• Powerful distributed cache - multiple clients, each client’s updates can be accessed by all other clients
• Effortless integration using Spring

Step 2: Scaling Up Your Database by Adding a Data Grid

If you need to scale up your database, you can use GigaSpaces as an In-Memory Data Grid. Your application communicates with the Data Grid using The GigaSpace Interface or the Map API interfaces. On the back-end, GigaSpaces persists the data to your existing database using your existing Hibernate O/R mapping in asynchronous manner.

Benefits:

• Database scalability through partitioning and data distribution - enables higher data volumes and higher throughput with low latency
• Better decoupling between your application and the database - no need to hard-wire Hibernate and database concepts into your code and runtime environment
• Event-driven model enables notifications when data is modified
• Database access can be synchronous or asynchronous - the GigaSpaces Mirror Service allows data to be persisted to the database asynchronously, without a performance penalty

Beyond the Data Grid: Scaling Out Your Application Using Space-Based Architecture

To gain ultimate scalability for your application, you can package your business logic together with the GigaSpaces space - which provides not only data but also messaging functionality. This creates a Processing Unit which is completely self sufficient (not dependent on a database, messaging server or any central component), meaning that you to duplicate it as many times as necessary without increasing complexity.

This approach is called Space Based Architecture, and the way to achieve it is the Spring-based OpenSpaces platform, which allows you to deploy your application as a set of services encased in a Processing Unit.

How to Switch Your Cache Provider to GigaSpaces and Gain Distributed Caching

This section provides instructions on switching your existing Hibernate local cache, typically EhCache, with GigaSpaces. This will give you a powerful distributed cache with multiple clients and advanced clustering.

To switch your Hibernate cache provider to GigaSpaces and use GigaSpaces distributed caching:

1. Download the GigaSpaces Download Page
   
2. Install GigaSpaces. If you need help, refer to the installation instructions.
   
3. This step is different if you are managing your Hibernate configuration using Spring (refer to the Spring documentation, ORM Data Access, section 12.2.2), or directly using hibernate.properties or hibernate.cfg.xml. Select the relevant tab below.

<prop key="hibernate.cache.provider_class">
    org.openspaces.hibernate.cache.SimpleMapCacheProvider</prop>

<prop key="hibernate.cache.use_second_level_cache">true</prop>

<prop key="gigaspace.hibernate.cache.url">/./dataGrid</prop>

hibernate.cache.provider_class org.openspaces.hibernate.cache.SimpleMapCacheProvider

hibernate.cache.use_second_level_cache true

gigaspace.hibernate.cache.url /./dataGrid

<property name="hibernate.cache.provider_class">
    org.openspaces.hibernate.cache.SimpleMapCacheProvider</property>

<property name="hibernate.cache.use_second_level_cache">true</property>

<property name="gigaspace.hibernate.cache.url">/./dataGrid</property>

1. Set or add the cache usage (the cache concurrency strategy) in your mapping resource file (*.hbm.xml files) to read-only, read-write or nonstrict-read-write:

<class name="your.class.name" table="YourTableName">
     <cache usage="nonstrict-read-write"/>

Alternatively, use the <class-cache> and <collection-cache> elements in your hibernate configuration file.

You must include the following JARs from the Hibernate distribution package:

jdbc2_0-stdext.jar; dom4j-XXX.jar; commons-collections-XXX.jar; asm.jar; asm-attrs.jar;
antlr-XXX.jar; ejb3-persistence.jar; hibernate3.jar; hibernate-annotations.jar;
hibernate-commons-annotations.jar

1. Since your Hibernate objects and their custom type will be eventually be serialized across to wire stored on remote space or replicated into a backup space In-Memoery, these need to be serialized. You should make all your Hibernate objects and their custom fields will be implementing Serializable.
2. Launch your application. The application should instantiate a local cache based on an embedded GigaSpaces space.

1. If you want to use GigaSpaces as a distributed cache, select the distributed topology that interests you from the tabs below.

Basic Guidelines

• For the application components that are read/write (up to 50% write/remove/update) you should have the hibernate 2nd local cache to use a remote partitioned space topology.
• For the application components that are read mostly (above 50% read) you should have the hibernate 2nd local cache to use a master local cache/local view with remote partitioned space topology.
• The embedded replicated cache topology is recommended only when having few client applications (up to 3) where the total amount of data can be accommodated within a single JVM and the application is 50% read/write.
• For the application components that are write mostly (above 50% write/remove/update), you should not use hibernate API directly , but move into the GigaSpace Interface and use the Mirror service (if you still want to persist the data). In this case you can leverage SQLQuery or Task Executors for complex distributed activities. In this case you should use remote partitioned space topology.

Settings GigaSpaces local cache with Hibernate Second Level Cache

When using Hibernate Second Level Cache you can configure also GigaSpaces local cache. See below an example for both:

<property name="hibernateProperties">
   <props>
    <prop key="hibernate.dialect">org.hibernate.dialect.${jdbc.platform}Dialect</prop>
    <prop key="hibernate.cache.provider_class">${hibernate.cache.provider_class}</prop>
    <prop key="hibernate.show_sql">{hibernate.show_sql}</prop>
    <prop key="hibernate.jdbc.batch_size">${hibernate.jdbc.batch_size}</prop>
    <prop key="hibernate.cache.use_query_cache">true</prop>
    <prop key="hibernate.transaction.flush_before_completion">true</prop>
    <prop key="hibernate.generate_statistics">true</prop>
    <prop key="hibernate.cache.user_structured_entries">true</prop>
    <prop key="hibernate.cache.use_query_cache">true</prop>
    <prop key="hibernate.transaction.auto_close_session">true</prop>
    <prop key="hibernate.connection.release_mode">${hibernate.connection.release_mode}</prop>
    <prop key="hibernate.cache.use_second_level_cache">true</prop>
    <prop key="gigaspace.hibernate.cache.url">jini://*/*/dataGrid?useLocalCache</prop>

    <prop key="space-config.engine.cache_size">5000000</prop>
    <prop key="space-config.engine.memory_usage.high_watermark_percentage">65</prop>
    <prop key="space-config.engine.memory_usage.write_only_block_percentage">63</prop>
    <prop key="space-config.engine.memory_usage.write_only_check_percentage">60</prop>
    <prop key="space-config.engine.memory_usage.low_watermark_percentage">55</prop>
    <prop key="space-config.engine.memory_usage.eviction_batch_size">1000</prop>
    <prop key="space-config.engine.memory_usage.retry_yield_time">100</prop>
    <prop key="space-config.engine.memory_usage.retry_count">20</prop>
   </props>
</property>

Custom Hibernate Second Level Cache Implementation

You may implement your own custom Hibernate Second Level Cache by extending the AbstractMapCacheProvider and implementing the buildCache method. The implemented class name should be set as the hibernate.cache.provider_class value. See below example how you may create a custom Hibernate Second Level Cache implementation that will have a different TTL for each cache region:

public class CustomMapCacheProvider extends AbstractMapCacheProvider {
    public Cache buildCache(String regionName, Properties properties)  throws CacheException {
    Long ttl = 0l;
    if (regionName.equals(Person.class.getName())) {
        ttl = 60000l; //TODO: make this configurable
    } else if (regionName.equals(Car.class.getName())) {
        ttl = 120000l;
    } else if (regionName.equals(House.class.getName())) {
        ttl = 180000l;
    } else {
        ttl = getTimeToLive();
    }
    return new SimpleMapCache(regionName, getMap(), ttl, getWaitForResponse());
    }
}

GigaSpaces Hibernate 2nd Level Cache Properties

GigaSpaces Hibernate 2nd Level Cache controlled via the following system properties: