If you want a major performance boost out of your application, then caching should be a part of your strategy. No doubt you have experienced moments in coding where you needed to store sets of data so that you don’t have to go back to the source to get them every time. The simplest form of caching is lazy loading where you actually create the objects the first time in memory and from there on out, you access them from memory. In reality, caching gets a lot more difficult and has many considerations.
- How do I cache in a distributed environment
- How do I expire items in the cache
- How do I prevent my cache from overrunning memory
- How do I make my cache thread-safe and segment it
All of these are concerns that you will have if you “roll your own” solution to caching. Let’s just leave the heavy lifting to the Spring Framework and we can go back to concerning ourselves with solving the complex problems of our domain.
Spring has a caching mechanism/abstraction for both Java and .NET, although the Java version is far more robust. Caching in Spring is accomplished through AOP or Aspect Oriented Programming. A caching annotation (Java) or attribute (.NET) can be placed on a method or a class to indicate that it should be cached, which cache should be used and how long to keep the resources before eviction.
Java Spring Cache with EHCache
In Java, caching with Spring couldn’t be easier. Spring supports several different caching implementations but EHCache is the default and by far my favorite. EHCache is robust, configurable and handles a distributed environment with ease. Let’s look at how we can add the ability to cache to a Spring project.
Application-context.xml
<beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:cache="http://www.springframework.org/schema/cache"
xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsdhttp://www.springframework.org/schema/cache http://www.springframework.org/schema/cache/spring-cache.xsd">
<cache:annotation-driven />
<!-- Ehcache library setup -->
<bean id="ehcache" class="org.springframework.cache.ehcache.EhCacheManagerFactoryBean" p:config-location="ehcache.xml"/>
Now that we have our cache setup, we can start to utilize it.
@Cacheable(name="records", key="recordList") //Cache the output of records, return it if already cached
public Collection findRecords(RecordId recordId) {...}
@Cacheable(name="records", key="recordsList", condition="recordType == 2") //Only cache if record type is 2
public Collection findRecords(int recordType) {...}
@CacheEvict(value = "records", allEntries=true) //Reload the cache, evict all entries
public void loadAllRecords() {...}
In the above example, we specify through the annotation that the records collection will be stored in a cache named “records” and the key to access the collection will be called “recordList”. The key parameter is optional. We also displayed an example of using Spring Expression language to process the cache conditionally. Remember that caches are defined either dynamically like above or in the ehcache.xml. For most complex caching scenarios, you will want to define the cache in the ehcache.xml with eviction and distribution rules and Spring will find it by the name parameter in the annotation.
What about .NET?
In .NET, you have a very similar mechanism to managing a cache.
<!-- Apply aspects to DAOs -->
[CacheResult("MyCache", "'Record.RecordId=' + #id", TimeToLive = "0:1:0")]
public Collection GetRecord(long RecordId)
{
// implementation not shown...
}
Remember that with the .NET cache, you provide the caching implementation in the XML just as you do in the Java version. In this example, we have used the provided AspNetCache.
What about controlling the cache yourself, querying the cache and more complex operations? Well even that is simple by merely autowiring the cache class or retrieving it from the context.
@Autowired
EhcacheCacheManager cacheManager;
...
EHCache cache = cacheManager.getCache("records");
Collection records = cache.get("recordList");
Built-in .NET Caching
Fortunately for .NET users, there is also a built in caching framework already in .NET. This technique is used mostly in the MVC and ASP world and I am not particularly fond of it since it is specifically geared for the web side of an application. I would prefer it to be more generic like the Spring solution, but it also has some advantages such as you can configure the caches in the web.config, you can also create custom caching implementations that utilize different mechanisms for cache management. Here is a great example utilizing MongoDB. The .NET version of the cache works much the same way as the Spring one does. Here are some examples with configuration.
[OutputCache(CacheProfile="RecordList", Duration=20, VaryByParam="recordType")]
Public ActionResult GetRecordList(string recordType) {
}
Now the configuration in the web.config…
//web.config
<caching>
<outputCacheSetting>
<outputCacheProfile>
<add name="RecordList" duration="3600" />
</outputCacheProfile>
</outputCacheSetting>
</caching>
If we are deploying our application to the cloud in Azure, we can use the AppFabric cache as demonstrated here.
Hibernate, Data Nucleus, JPA Result Caching
Another thing to keep in mind is that when you are using tools such as Hibernate, caching is built in with these solutions. There is a built-in second level cache implementation that is provided when you configure Hibernate and I tend to use EHCache for this as well. You must remember to add the Hibernate Cache annotation onto your objects at the class level that you want to cache as well. A properly setup ORM solution with Spring and Hibernate with a properly configured second-level cache is very hard to beat in performance and maintainability.
Conclusion
We have really done a lot with very little thanks to Spring and caching. While caching is powerful and will help improve the performance of your application, if it is overdone, it can cause problems that are difficult to diagnose. If you are caching, make sure you always understand the implications of the data and what will happen throughout the caches lifecycle.
Over the years, I’ve had to learn many programming languages in the course of my career; Perl, JavaScript, shell, Java, C#, Objective C to name the most prominent. Several years ago, I wrote a little Python for a small project and I hated the language for some of the syntactical eccentricities that makes it unique from other languages out there. Lately, I had to pick up Python again, just to be pleasantly surprised that I can actually enjoy it now.
