| Server Role | Number of servers per unit | Processor | Memory | Operating system | Disks* |
| Operational database server | Two per management group—Windows Clustering cluster | Two dual core | 8 GB | Windows Server 2003 with SP1, 64 bit | SAN Windows Clustering Quorum: 5 GB SQL data: 130 GB RAID 10 SQL log: 10 GB RAID 5 TempDB: 10 GB RAID 5 |
| Root management server | Two per management group—Windows Clustering cluster | Two dual core | 16 GB | Windows Server 2003 with SP1, 64 bit | SAN Windows Clustering Root management server state drive: 5 GB |
| Failover database server | One per management group | One dual core | 8 GB | Windows Server 2003 with SP1, 64 bit | SAN SQL data: 130 GB RAID 10 SQL log: 10 GB RAID 5 TempDB: 10 GB RAID 5 |
| Operations Manager 2007 Audit Collection database | One per Operations Manager 2007 Audit Collection deployment | Two dual core | 8 GB | Windows Server 2003 with SP1, 64 bit | SAN SQL data: 1 terabyte RAID 10 SQL log: 50 GB RAID 5 TempDB |
| Management server | Two per management group | One dual core | 4 GB | Windows Server 2003 with SP1, 64 bit | Not applicable |
| Management Server and Operations Manager 2007 Audit Collection | One per Operations Manager 2007 Audit Collection deployment | One dual core | 4 GB | Windows Server 2003 with SP1, 64 bit | Not applicable |
| Data warehouse | One per the entire deployment | Four dual core | 16 GB | Windows Server 2003 with SP1, 64 bit | SAN SQL data: 400 GB RAID 10 SQL log: 30 GB RAID 10 TempDB: 30 GB RAID 10 |
| Gateway server | Two per security-enhanced environment | Two dual core | 4 GB | Windows Server 2003 with SP1, 64 bit | Not applicable |
| Role | Redundancy/high availability | Disaster recovery |
| Management server | Approach: Deploy multiple management servers for agent failover. Description: In a management group, if one management server goes offline, the agents and gateway systems reporting to it will fail over to another management server in the management group. In Microsoft IT, agent failover behavior is defined in Active Directory, where agents can automatically discover it. | Approach: Deploy management servers in multiple locations. Description: If one site goes offline, the agent will fail over to the management server in another site, assuming that the site’s failover configuration allows this. Microsoft IT is considering having management servers hosted on virtual machines in the failover location, as opposed to procuring more dedicated hardware. |
| Root management server | Approach: Cluster the root management server. Description: By using Windows Clustering, Microsoft IT has deployed the root management server role onto a two-node cluster to allow the underlying servers to go offline without having to take the role itself offline for a prolonged amount of time. | Approach: Promote a remote management server to root management server. Description: If the root manager server cluster becomes unavailable, Microsoft IT will promote one of its remaining management servers into the root management server role. Management servers are deployed in remote locations to account for a site-wide outage. |
| Gateway server | Approach: Deploy multiple gateway servers per location. Description: The approach is the same as management servers, with two notable exceptions. First, multiple gateway servers need to be deployed in each distinct network/Active Directory space where the gateway’s agents are located. Second, gateway servers need to have multiple management servers that they can communicate with to ensure that they themselves can fail over if necessary. | None: Because the gateway servers are located in the same physical segment as the agents, if the location is offline, the agents will be offline as well. |
| Operational database server | Approach: Cluster the SQL Server–based server that the Operational database is hosted on. Description: By using Windows Clustering, Microsoft IT has deployed the SQL Server–based server that the Operational database role is running on onto a two-node cluster to allow the underlying servers to go offline without having to take the role itself offline for a prolonged amount of time. | Approach: Implement log shipping to a remote SQL Server–based server. Description: A stand-alone SQL Server–based server has been deployed in a remote location and log shipping was set up to replicate data from the primary Operational database to the failover SQL Server–based server. If a disaster occurs, the management group can be manually pointed to use the failover database. |
| Operations Manager 2007 Audit Collection collector | None. | None. |
| Operations Manager 2007 Audit Collection database | None. | None. |
| Data warehouse | None. | Approach: Use database backups. Description: Scheduled periodic database backups with off-site storage are implemented. If a disaster occurs, the latest backup sequence is restored to a new server. |
