Why OSCache Support is not there yet?

[kaleidoUser01]

Hi Team,

Thank you very much for the kaleido bundle.

Currently I am thinking to use CacheManager of  kaleido cache. While going through the kaleido CacheManager  implementation, I realized that no cache manager implementation for OSCache yet. Any reason?

Also, to explain the benefits of CacheManager (https://code.google.com/p/kaleido-foundry/wiki/CacheManagerExample), it is mentioned that 

Benefits 5 - " You can still use specific functionalities of your cache provider," what does this mean?

Also,keen to know if any performance impact & CPU resource utilization,memory foot print of using kaleido CacheManager?

What is the road map for this development? Any group is using it in there production ?

How could we integrate CacheManager with Hibernate?

Finally, as a developer how can I contribute to this project?

Thanks & regards,

Abdus Mondal

[jraduget]

Hi Abdus,

Interesting... I didn't know OSCache before. I have no need to integrate it for my customers, that's the reason why it is not implemented... 

Recently we integrate Apache JCS, this is quite simple because lot of work are done on our abstract layer (also for the unit tests)

I created the issue here, to integrate OSCache :  https://code.google.com/p/kaleido-foundry/issues/detail?id=41

And reference OSCache in the wiki : https://code.google.com/p/kaleido-foundry/wiki/CacheManagerExample

For the benefits 5, when you handle a kaleido cache / cacheManager, you still can access to the underlying cache using #getDelegate()

http://kaleido-foundry.googlecode.com/svn/tags/kaleido-0.9.0/site/apidocs/org/kaleidofoundry/core/cache/CacheManager.html#getDelegate()

This is not recommended because you loose the portability, but this is possible...

I did some performance benchmarks of "kaleido CacheManager" at some of my customers. The result was very specific to their application (intensive use of caching). Results were quite good (less than 1% of cpu overhead using kaleido, and no real impact on average response time). The interesting thing is that with kaleido, we can quick switch and compare cache solutions (ehcache, jboss cache, ...).  Obviously, the results depends a lot of the cache configuration and topology that have been chosen.

We didn't go for the moment in the way to integrate CacheManager with hibernate, because the first goal of this API was to manage directly the caching mechanism in the repository / service layer. I confess that I had very bad experience and bottleneck with the hibernate cache (second level)... memory consummation, cpu overhead because of the serialization mechanism (with ehcache)...

Glad to see that you are interesting to contribute :D You can clone the project from github, or checkout from svn, and provides your contributions as patches or pull request.

And if you are motivated, why not becoming a project member.

Best regards,

--

Jérôme R.

[kaleidoUser01]

Thank you Jérôme for the quick reply.

Let  me enhance the kaleido API knowledge (which I am doing day by day ;) )and parally I will note down all the findings/concern/  I have. 

I will keep you posting on it.

Cheers!

Thanks & regards,

Abdus