Hi Guys,
         I am having an object(Hashtable), if I make any change in the object it should reflect in all the instances in my clustering environment.
Is it possible to do with JNDI?.

Please clear my doubt.

thanks,bala

Hi Bala,

There are clustering environments that do this, some that don't. The trick is spotting the change. It depends of your clustering environment just updates the data or replicates method calls over the network to remote objects.

I'm sure Cameron will enlighten you on one particular product ;-)

What are you using?

Regards,
Max

Hi Max,
             Right now I am using Weblogic Server 7.0.
I don't want to use any EJB's here.

thanks,bala

Max: "There are clustering environments that do this, some that don't. The trick is spotting the change. It depends of your clustering environment just updates the data or replicates method calls over the network to remote objects. I'm sure Cameron will enlighten you on one particular product ;-)"

Never. Not me. No siree. Just because he could use the clustered singleton pattern, or some form of JCache-based replicated cache ... well, I've probably already said too much ;-)

p.s. See sig below for a suggestion.

Peace,

Cameron Purdy
Tangosol, Inc.
Coherence: Easily share live data across a cluster!

You can use TCP/UDP multicasting to synchronize the hashtable.
Puru

Yes, but,

a) unless you use Pragmatic General Multicast Whats PGM? you can't guaranteed packets wind up anywhere or that anything is in a consistant state.

b) You have to write these wrappers yourself, I did, Coherence (yes Cameron;-) was a bit quicker and they can very well re-write stuff to use PGM (other clever stuff) when it's in full swing, not I.

Regards,
Max
p.s. How would you implement a monitor to check data has changed? It's an interesting question once you delve into it, have a go :-) I found dynamic proxies useful.

"Pragmatic General Multicast"? Did you know that it actually stands for "Pretty Good Multicast"? ;-)

Tangosol Coherence isn't bound to Multicast, and thus PGM isn't particularly applicable. Coherence uses Datagram Multicast when it has a compelling reason to, but generally sticks to unicast (the UDP protocol) if it can. That's a lot more scalable on a switched network. (Obviously, when it does replicated caching, it will often use multicast to communicate updates, since that is the most efficient approach with more than two nodes.)

Have you had a chance to try Coherence?

Peace,

Cameron Purdy
Tangosol, Inc.
Coherence: Easily share live data across a cluster!