[mule-user] Mule Clustering
Vadim Shatokhin
We‘re investigating MuleSoft enterprise products to build an enterprise distributed platform for our customer. The Mule software has been chosen due to wide integration abilities and flexibility. But questions around clustering in the Mule are still unclear for us. Could you answer the following questions?
We need to have a hierarchy of services like root -> domain -> server. Each domain contains one or more clusters of nodes for the load balancing and performance.
1. Mule Cluster.
We’ve seen that all Mule server nodes have ID, domain ID and Cluster ID but haven’t seen any reference where they are really used except maybe monitoring. Does the Mule provide ready solution to support and governance nodes in a cluster to provide, for example, right routing when some of nodes are turned off?
2. Galaxy Cluster
Could the Galaxy register be clustered? That is, when register database should be distributed between several data centers.
3. Load Balancing.
Does the Mule provide load balancing or no?
4. Bi-direction relationship between Mule Server Nodes and Repository/Registry (Galaxy) to provide right routing when some of nodes are turned off.
Thanks
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
Andrew Perepelytsya
1. Mule Cluster.
We’ve seen that all Mule server nodes have ID, domain ID and Cluster ID but haven’t seen any reference where they are really used except maybe monitoring. Does the Mule provide ready solution to support and governance nodes in a cluster to provide, for example, right routing when some of nodes are turned off?
2. Galaxy Cluster
Could the Galaxy register be clustered? That is, when register database should be distributed between several data centers.
Yes.
3. Load Balancing.
Does the Mule provide load balancing or no?
Load balancing is tightly coupled to your architecture and very much depends on transports used. It can be done either internally by Mule or on a higher level outside of Mule process. Just understand that there's no one-size-fits-all flip-switch solution.
4. Bi-direction relationship between Mule Server Nodes and Repository/Registry (Galaxy) to provide right routing when some of nodes are turned off.
HTH,
Andrew
Vadim Shatokhin
1. Mule Cluster.
>http://www.mulesoft.org/display/MULE2USER/Mule+High+Availability
HA realizes Active-Passive schema, but we need Active-Active with direct node addressing (Star). As I understand Mule doesn't provide such solution?
2. Galaxy Cluster
>>Could the Galaxy register be clustered? That is, when register database should be distributed between several data centers.
>Yes.
Is it providing via DB changing functionality providing in WAR configuring in enterprise version or other?
3. Load Balancing.
>>Does the Mule provide load balancing or no?
>... It can be done either internally by Mule or on a higher level outside of Mule process. ...
That is, on routing we need check available nodes via Galaxy. Is it right? Could you provide a references to description or example of internal load balancing? As I understand, load balancing is a one of major functionality of routing and ESB, or not?
4. Bi-direction relationship between Mule Server Nodes and Repository/Registry (Galaxy) to provide right routing when some of nodes are turned off.
>How is that different from load balancing above?
The Galaxy as registry should provide a service life cycle management and takes bi-directional relationship with each of service to know about its current state. When we need send direct request to particular node it is preferable to know about its availability before the sending to have a possibility, for example, send a message to special queue, or add exception handling. I haven't found any examples about