Hello, I have a question about best practice for SCCM in a HA datacenter. We have two locations, our primary office, and our colo datacenter (which is unmanned) connected with a high speed link. Currently, our central site is a primary site in our central office and the colo datacenter is a secondary site in another city. If either the central site goes down, or the link between sites is temporarily severed, we need the secondary site to remain functional (especially the OSD and SUP pieces). Currently, this is not possible in our setup. Would the best thing to do here be to simply, remove the secondary site at the colo datacenter and replace it with a primary site that is a child to the central site in the primary office? At that point, we would manage everything from the central site and it would replicate down? If the central site went down briefly, would this have any immediate impact on the child site? Also, across a high speed link (50MB), would changes like creating a new advertisement, or a new small package, etc. happen very quickly in this setup (would it would send the changes immediantly and just take however long it takes to copy the bits?). Is this the best plan, or anyone else have a better idea? Thanks!

For the most part your biggest issue will be a backlog of HW&SW inventory files! I have servers down for almost 3 weeks and it took a few days to clear out the backlog. That is the biggest issue you will have. http://www.enhansoft.com/

Thanks for the replies! If I understand correctly, yes Garth that is right. If the Central Site (Site A) goes down, I want Site B (child primary site) to continue to function. I don't care about the clients in Site A staying up (at this point), as long as site B still functions I am okay. Site A and Site B are completely different subnets, no boundaries should not be a problem (no overlapping). Clients in Site A (ie. 192.168.150.10) will get picked up by the boundary for Site A (Central Site) (192.168.150.0/24) and Clients in Site B (ie. 192168.200.10) will get picked up by boundaries for site B (192.168.200.0/24). Clients will not roam between sites, but only be up when their primary site is up. That make sense? Ultimately I would be nice to have full redundency, but I am aware of the limitations of that with the current version of the production (hopefully vNext will provide a solution here for that). Some backlog issues are okay, shouldn't cause me any major problems. I am okay with not being able to get NEW SU, as long as I can deploy my current OSDs with current SU's. Sounds like what I need to achieve is doable with your solution. Thanks!

That makes sense (our biggest goal currently is actually to solve for the link going down for an extended period more than the server itself going down, so that isn't as applicable right now, but does provide some good future options). Hopefully vNext will provide some additional features in this area. Thanks!