Step C2: Create the Site Link Design
Published: February 25, 2008
Site links are used to connect the defined sites in Active Directory. The site links reflect the inter-site connectivity and method used to transfer replication traffic. All sites must be connected with site links if the domain controllers at each site are to replicate. By default, all sites belong to the default-first-site-link, with replication scheduled to occur every 180 minutes, each day of the week. Site links can be created at any time and sites can be added or removed easily. However, there could be an impact on replication because of latency issues when reconfiguring sites, site links, and scheduling associated with the site links. Task 1: Determine the Site Link DesignActive Directory automatically creates the default-first-site-link. When all sites in the design are connected and have the same connectivity and availability to one another, a single site link can be used to represent the links between the sites. This full mesh design assumes that all sites are well connected and that there is no need to design specific links between sites. This approach simplifies the design by eliminating the need to design site links, as well as automatically configuring the site link structure. Since the connectivity and availability of the links are identical, the replication schedule, interval, and cost will be configured identically. This choice is only useful when all of the sites are connected by WAN connections with identical available bandwidth and latency. If sites are connected with physical network links that have different costs of usage, availability, speed, or available bandwidth, there may be a need for different replication schedules. A new site link would need to be created to account for these differences. Site links use a cost algorithm to influence which path replication traffic will use to flow between sites. A preferred connection would be configured at a lower cost than a less-preferred connection. The replication system uses the link with the lowest cost. If there is a dollar cost to using a link, the link might be assigned a higher cost value as well. The replication of traffic across the link is controlled by the availability schedule and how frequently the link is set to replicate. For example, a link can be configured to replicate every 30 minutes during the hours of 2:00 A.M. to 4:00 A.M., Monday through Friday. Site scheduling can specify intervals as brief as every 15 minutes, ranges from any time of the 24 hour clock, and any combination of days of the week. When assigning the replication schedules and intervals, care should be taken to ensure that any replication goals required by the organization are met. Replication goals can be defined such that all changes are recorded in a set period of time for the following:
When defining the replication schedules and intervals, ensure that all replication goals are met for worst case scenarios. That is, can a change originating in one site replicate within the time frame with the site that is the greatest number of hops away from the originating site? If it is not possible to meet the goal, then the interval and schedule need to be updated or the goal needs to be redefined. Consider a topology that consists of five sites (A–F), consisting of a single forest with four domains. The sites are connected to one another through direct site links with the replication schedule configured for 24 hours and the interval set at the default of 3 hours. For the purposes of this example, there are two replication goals: one for schema and configuration convergence to be completed in 6 hours and one for global catalog convergence in 4 hours.
Figure 4. Configuration and schema convergence Because of the number of hops involved, it is not possible for a change introduced in site E or F to converge at site C within 6 hours; it would take a minimum of 9 hours (3 hops). Either the replication goal would need to be updated or the interval would need to be set to another value.
Figure 5. Global catalog convergence In the example of global catalog convergence, there are four global catalog servers in sites A, B, D, and E. If a change is introduced in site D, then all sites will be updated within the 4-hour replication goal. However, if the change is introduced in sites A, B, or E, then it will not be possible to meet the goal of 4 hours as it will take a minimum of 6 hours (2 hops) to reach all sites. Again, the replication goals would need to be updated or the interval time frame changed on the site links. Associate all sites with similar links to the new link and remove the sites from the default-first-site-link. For each site link identified, record the name of the site link, the cost associated with using the link, and the link’s replication schedule and interval. For each site, record the link that is used to connect it to other sites. A site may have multiple site links associated with it. Decision SummaryLinks between all sites should be defined through the use of one or more site links. If any sites are disconnected from the others, the Knowledge Consistency Checker (KCC) will generate an error message. The site links control the replication of the directory database between domain controllers in different sites and, if multiple paths are available, control which path is preferred. The site link design can be changed. However, changing the site links may have an impact on the performance of directory changes until all updates converge. Additional Reading“Creating a Site Link Bridge Design” at https://technet2.microsoft.com/windowsserver/en/library/5d05f4ed-a9ec-4dac-b9a8-8527b6c8e0da1033.mspx?mfr=true |
|
.jpg)
.jpg)
.jpg)