Usage

NitroCache implements java.util.Map interface, which makes it extremely easy to use and pluggable.

Creation

NitroCache<String,String> _cache = null;
// Every call to NitroCache.getInstance will create new cache. Its not singleton
_cache = NitroCache.getInstance(5000, CacheEviction.FIFO); 

Inserting value

_cache.put("1000","hello world");
System.out.println(_cache.get("1000");

Removing key

_cache.remove("1000");

Method list

Method	Description
clear()	Clears the cache. Internally creates a new cache
containsKey(Object)	Returns true if key present
containsValue(Object)	Not Supported
entrySet()	Return an entrySet based on current entries*
get(Object)	Return value if key is present
isEmpty()	Return true if Cache size > 0*
keySet()	Return set of keys from the cache*
put(K, V)	Add key/value to cache
putAll(Map<? extends K, ? extends V>)	In a loop add all key/values
remove(Object)	Remove the key if present
shutdown()	Stop and clear underlying cache
size()	Return size of cache*
values()	Return all values from cache*

• Methods marked with "*" create a secondary map and use it for operation. The in-flight entries may/may not show-up as this method does not add any lock on the cache, thus avoiding performance hit

Eviction Modes

Mode	Description	Suggested Usage
FIFO	Standard First-in-first-out algorithm. put(key,null) will remove the entry and compacts cache. Adding same key/value again moves the entry to end.	Frequent removal and re-entry of key/values
FIFO_V2	Optimized for speed FIFO. put(key,null) removes entry but does not immediately compact cache. Adding same key/value again does not change order of entry in FIFO queue	Less frequent removal/replace of values for keys. High throughput.
LRU	Standard Least Recently Used algorithm	If LRU provides more hits

[Shay Hassidim]

Throughput results are very deceiving. These does not tells the real story.
Stable latency response time over a long period of time is what you need to measure. This is what you need from a caching product.
I suggest you take a look on these:
http://www.gigaspaces.com/benchmarks
Can any of these products deliver the same latency with such outliers distribution?
See more here:
http://blog.gigaspaces.com/possible-impossibility-the-race-to-zero-latency/

Shay