A Concurrent Result Cache solution adopted from the code sample in the book: Java Concurrency in Practice by Brian Goetz.
There are occasions that an expensive operation needs to be ran in a multi-thread environment. The following code is a solution to be used to avoid running the expensive operations more than once.
The Interface of expensive computation:
|
public interface Computable<A,
V> {
public V compute(final A arg) throws
InterruptedException;
}
|
The concurrent result cache implementation:
This implementation guarantees that only one expensive
execution will be invoked, other thread will be waiting for the Future to
return.
|
public class ConcurrentResultCache<A,
V> implements Computable<A,
V> {
private static final Logger logger = LoggerFactory.getLogger(ConcurrentResultCache.class);
private final ConcurrentMap<A,
Future<V>> cache = new ConcurrentHashMap<A, Future<V>>();
private final Computable<A,
V> c;
public
ConcurrentResultCache(final Computable<A, V> c) {
this.c = c;
}
public V compute(final A arg) throws
InterruptedException {
logger.trace("entered
compute...");
// waiting result
to be available. Intercept any CacellationException.
while (true) {
logger.trace("before first
get");
Future<V> f = cache.get(arg);
logger.trace("after first
get");
if (f == null) {
logger.trace("f is
null");
// Creating a new
task wrapped in a Callable.
Callable<V>
eval = new Callable<V>()
{
public V call() throws
InterruptedException {
return c.compute(arg);
}
};
FutureTask<V>
ft = new
FutureTask<V>(eval);
f = cache.putIfAbsent(arg, ft);
logger.trace("after
putIfAbsent()");
if (f == null) {
// if the task is
put into the cache first time, run the task
//
in the SAME thread.
f = ft;
logger.trace("before
run()");
ft.run();
}
}
try {
logger.trace("before
second get");
return f.get();
} catch
(CancellationException e) {
cache.remove(arg, f);
} catch (ExecutionException
e) {
throw new
IllegalStateException("Task execution exception caught.", e);
}
}
}
public void remove(final String key) {
this.cache.remove(key);
}
}
|
One example of expensive operation: Property File upload.
|
public class PropertyLoader implements
Computable<String, Properties> {
private static final Logger logger = LoggerFactory.getLogger(PropertyLoader.class);
@Override
public Properties compute(final String filePath) throws
InterruptedException {
logger.trace("entered
PropertyLoader compute()");
return readProperties(filePath);
}
private Properties
readProperties(final String filePath) {
final String propFile = System.getProperty(filePath);
final Properties properties = new Properties();
try {
properties.load(new FileReader(propFile));
} catch (IOException e) {
logger.error("Failed to
load properties.", e);
}
return properties;
}
}
|
A usage example:
|
public class
PropertyUpdateManager {
private static final Logger logger = LoggerFactory.getLogger(PropertyUpdateManager.class);
private static final String ENV_PROPERTIES_FILE = "envPropertiesFile";
private static
ConcurrentResultCache<String, Properties> propertiesCache = new
ConcurrentResultCache<String, Properties>(
new PropertyLoader());
private static
PropertyUpdateConfiguration propertyUpdateConfiguration = null;
private static boolean cacheCleared = false;
public static Properties getPropertyMap()
{
if (propertyUpdateConfiguration == null || isTimeToUpdate(propertyUpdateConfiguration)) {
try {
clearPropertiesCache(ENV_PROPERTIES_FILE);
cacheCleared = false;
Properties properties = propertiesCache.compute(DFP_ENV_PROPERTIES_FILE);
} catch
(InterruptedException e) {
logger.warn("properties
update has been interrupted.", e);
properties = new Properties();
}
}
return Properties;
}
public static synchronized void
clearPropertiesCache(final String key) {
if (!cacheCleared) {
if (propertiesCache != null) {
propertiesCache.remove(key);
}
cacheCleared = true;
}
}
private static boolean isTimeToUpdate(final
PropertyUpdateConfiguration original) {
final long currentDt = System.currentTimeMillis();
final long timeSinceLastUpdate = currentDt - original.getLastUpdateDt();
logger.debug("timeSincelastUpdate:{},
updateDurationMs:{}", timeSinceLastUpdate,
original.getUpdateDurationMs());
return timeSinceLastUpdate > original.getUpdateDurationMs();
}
}
|
No comments:
Post a Comment