[ale] Extended or Global Clustering

[Tim Meanor]

Look at Veritas cluster server (VCS) and volume replicator (VVR).   
These allow you to build a cluster in which the nodes are  
geographically separated (ie a global cluster).  VCS handles the  
clustering duties, and VVR is used to synchronize data between the  
geographically separated nodes.  For example, with VCS you build a  
local cluster using shared storage between the nodes, and a remote  
cluster with the same configuration. Then you set up VVR to  
synchronize data from the local cluster to the remote cluster.  Then  
you use VCS to create a global cluster consisting of the local and  
remote clusters.  Once that is all done, you can fail over packages  
from a node in the local cluster to one in the remote cluster (and  
vice versa).  I'm in the process of setting this up where the local  
and remote clusters are separated by about 500 miles.



-Tim

On Nov 21, 2007, at 4:14 PM, Kevin O'Neill Stoll wrote:

> Hey everyone,
>
> I'm searching for whitepapers, books and/or people resources to  
> learn more about extended or global clusters. I'm more interested in  
> the "meat and potatoes" of the handful of ways people have  
> accomplished this. I know Weather Channel Interactive and Dell.com  
> offer this type of clustering model, but they are under NDA  
> agreements - so no dice.
>
> In my search I have found many books and resources that like to talk  
> about datacenter (layer 1 and 2) type issues for managing a cluster  
> (automation of bulids, config and patch management), discussions  
> about commodity hardware versus "big iron" or the like and various  
> product choices. I suppose what I'm hoping for is practical "hands- 
> on" knowledge of how such a setup was done from each layer of the  
> puzzle ( network, storage, systems, load balancing, database,  
> application servers )
>
> Seems the 1st thing that comes to mind when you mention this is,  
> start with dark fibre and work up from there but I'm sure there are  
> other variations to accomplishing this goal. As opposed to actual  
> configurations, I guess I'm focusing on the high level architecture  
> choices - realizing you shouldn't forget to deal with individual  
> applications which respect to how intelligent they are or which  
> frameworks they use (think EJB or ESB models) but that's a whole  
> other can of worms.
>
> Thanks for the help and listening to me ramble.
>
> Kevin Stoll
> "Some people are like slinkies - they don't serve much purpose, but  
> they make you smile when you tumble them down the stairs"
>
>
>
>
>       
> ____________________________________________________________________________________
> Be a better pen pal.
> Text or chat with friends inside Yahoo! Mail. See how.  http://overview.mail.yahoo.com/
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale