Project

General

Profile

Scaling Tigase MUC

E A
Added about 4 years ago

Hello,

I am looking for a way to implement MUC in cluster mode.

We host our servers on Amazon and use their load balancer (round robin).

We currently have Tigase set up as a cluster (Auto discovery) with two nodes on m3.large aws instances. (We or on testing phase)

From what I've read, there are two options to implement MUC in cluster:

1)

http://docs.tigase.org/tigase-server/7.0.0/Administration_Guide/html/#_load_balancing_external_components_in_cluster_mode

Problem:

  • How do I scale this? If I need a few servers for MUC, looks like I will need to edit config file on all nodes every time I want to add another server.

2)

http://docs.tigase.org/tigase-server/7.0.0/Administration_Guide/html/#_virtual_components_for_the_cluster_mode

All nodes redirect to a specific node which hosts the MUC component.

Problem:

  • Single point of failure, is it possible to implement when you need to deploy several MUC servers?

  • How do I scale this? If I need a few servers for MUC, looks like I will need to edit config file on all nodes every time I want to add another server.

Given these issues - what is the best solution I can use to implement this?

Thanks!


Replies (4)

Avatar?id=6023&size=32x32

Added by Artur Hefczyc TigaseTeam about 4 years ago

There is also a third option: Tigase ACS - Advanced Clustering Strategy which includes also advanced clustering for MUC and PubSub. Please refer to the wiki pages for more details and let me know if you have questions about it.

Added by E A about 4 years ago

Thanks Artur,

Can you advise:

1) Where can I see pricing for MUC ACS / etc. ? I understand these modules are not open source.

2) Is there a way to achieve MUC Clustering while maintaining scalability without ACS?

Much appreciated

(1)
Avatar?id=6023&size=32x32

Added by Artur Hefczyc TigaseTeam about 4 years ago

ad. 1. Please send us a message (you can send me an email or use a contact form: http://www.tigase.net/contact with your company name and address and I will send you back formal quotation with all the pricing details

ad. 2. The ACS was designed to provide high scalability through symmetric clustering. In my opinion this is the best and most optimal way to get true scalability. Any other way is just a workaround with significant limitations. And I think you already know other ways, through the external components, virtual components, or you can deploy MUC on each cluster node with a different component name. These all would give you some scalability but with a lot of manual work and in case of a cluster node failure not quick way to recover.

(1)

Added by E A about 4 years ago

Will do.

Thanks

    (1-4/4)