I am working with a large company designing a multi-cluster, multi-site DCRUM setup. We are trying to stay within supports recommendations, with a little leeway. Obviously we will not know how it will function until implementation occurs. Will keep everyone in the loop on how it works. Attached is current design . This is a work in progress. new-rum-install-diagram-v3-scrubed-for-answerhubrw.pdf
I've just finished deploying a big environment for a customer.
31 CAS, 31 ADS, 6 AMD, broken up into 7 clusters.
Cluster 1, includes 1 CAS and ADS, used for reporting/DMI/master config control
Clusters 2-7 each include 5 CAS, and 5 ADS each, and get the data from 1 AMD
The 6 AMDs are fed from a packet broker which splits the traffic 6 ways (using session aware load balancing so each AMD always gets a cohereent/complete traffic stream).
The customer also wants 1 minute intervals (instead of the default 5), and some extra monitoring detail being captured (typically we only capture content pages, customer wants content and all images/scripts/style sheets etc.). It is very much a non-standard implementation.