powered by Jive Software

Distributed / fault-tolerant Openfire?


I have been looking for a distributed solution of Openfire, something that allows me to have 2 servers acting as one. The only thing I have found is ejabberd, but it seems this is the only advantage it offers and wasn’'t satisfactory at all.

Is there any option / solution / intention concerning this based in Openfire?

Thanks a lot.


We’‘re working very hard on this feature at the moment. However, it will be commercial (as previously announced), part of which is based on the fact that we’'re using a commercial clustering library.

Best Regards,



Would you mind telling us a little more about the solution you’'re developing? i.e. Is it a:

a) Active/Passive solution or will all nodes support users at the same time?

b) Designed so that the servers are close to each other/have good network connectivity, or will it allow a distributed model where potentially multiple servers across the globe connected using low bandwidth links could support the same domain?

At the moment, we have clustering of sorts (we run Openfire on a VMWare ESX Cluster where it can failover between physical hardware) however we’'re a multinational company and because we want to support a single domain name have our remote users connect to the same server which makes our network connections a potential source of failure.




All nodes will support users at the same time. We’‘re using Tangosol’'s Coherence product as the backbone: http://www.tangosol.com.

It should be possible to have physically distant cluster nodes, but everything depends on the amount of traffic that has to be sent back and forth – we won’‘t know that until we’‘re further along in the clustering implementation. Our first detailed report on the clustering work is probably a couple weeks out, but we’'ll keep everyone posted.