Is there a pattern with JBoss for this kind of problem ?

[samfra]

Hello,

I would like to use JBoss for personal reasons and I am trying to see if it is adapted to what I need.

Imagine the following :

I have a graph (not a tree, a graph but without any cycle) of "things".
I would like to put in memory all relations between the things
I would have for example

public class Thing {

private ArrayList myParents;
private ArrayList myChildren;
private Object my_property_n1;
private Object my_property_n2;
...
}

I have about 50,000 things.

How would be the best solution to use this in JBoss ?? can I do this with EJB's ??

I don't understand where I can put my static objects. 50,000 things is long to get in memory, so I have to do it just once !!

Thanks for helping and telling me where I can put this cache level ....

[mittalr]

hello,
well this seems to a problem of managing 50,000 Things object. I guess the simple solution will be to have a class containing hashmap or an array list based on your needs to store this "Things" object. and then make that class a singleton or a monostate which can access your Things obect in the way u want it. I don't think that you need any type of EJB for this.

I hope this helps

-Rajesh

[carpy1970]

The question you really need to ask is do you actually need your 50,000 "things" in memory, or can u just load the "things" u need as and when u need them. 50,000 "things" is going to take a long time to load and a lot of memory however you do it.

[skondam]

You can use VeryLargeHashtable library to store the things in memory or on the disk based on available memory
http://www.webmacro.org/VeryLargeHashtable

[aliv]

One way as carpy1970 said is to load thing only when you really need to access its properties. About relations, another way of relations representation is connectedness matrix (M), things with numbers n1 and n2 are related when M(n1,n2) = 1 otherwise M(n1,n2) = 0, so this matrix has 50000*50000 cells, every cell is one bit, size of such matrix is about 312.5 Mb, it is a bit much, but it is the fast way to surf between graph leafs.