Is it correct to have thousands of Nodes with only 1 element

[fatefree]

The data I am trying to cache is a simple string request and string response, completely dynamic and compressed down to about 200 characters each.

Initially I had these inside a single node, but I found out I could not expire them at the element level, but only the node level. Is it correct, meaning is the way JBoss intended to be used is to create new nodes for every of these requests? I will be storing anywhere from ten thousand to a hundred thousand nodes. Will this be too many nodes from a performance perspective?

[liuhang781102]

Node is the basic unit of treecache, so if you catch many data in a simple node, you are not using the facility of treecache . so ,thousands of nodes is needed.

[mircea.markus]

I don't expect any significant performance drawback; it's the memory footprint that will be bigger. Isn't it possible for you to perform benchmark? If so, please post your figures

[liuhang781102]

I didn't perform benchmark before. I am working for a enterprise application now . The system presure is mainly on the database. The application server is much flexable.

I think that The main adverange of treeCache is transactionable,more functions . If you just want a simple cache with LRU policy, you may use other cache, or write it your self.

[liuhang781102]

I suggest that you can encapsulate a cache interface of yourself. The main method is 'get' and 'set' ,then you can change the implementation as you want.