Does anyone have any experience with running ARS in a distributed architecture? I'm currently testing a viable solution and have a few questions.
1) I have run into an issue with syncing the History database. Our lab history database is 50gb and the initial sync is not complete before the next snapshot is created. This causes the sync to fail. History database on our production environment is 83gb. Is there a better way to do the initial sync? Can the database be backed up, copied across the WAN to the target SQL server, brought online, joined to the local ARS Server and then set as a sync database? Will any changes in the master DB be synchronized or will the file be re-created?
2) Due to number 1 I configured ARS to use Config DB in Publisher/Subscriber mode and History DB pointing to a single server. The issue with this solution is the MMC is unusable if the History DB is lost or unavailable. Is there a way to tell ARS to ignore the History DB? Or would a blank/fresh database be needed until the master is back online?
We have users spread across the globe and i'd like to split our ARS infrastructure across 3 continents. 1 x Americas, 1 x EMEA, 1 x APAC. Mainly for speed where I can point each local ARS server to it's closest SQL server, as well as redundancy in case of failures.