Home > Storage > Data Storage Essentials > Storage Resource Manager (SRM) > Dell SRM: Disaster Recovery and Federated Architecture > Introduction
Federated architecture considers the existence of a global data center, where time-series, events, and topology data from regional sites is replicated and global users can access the data transparently using their own set of reports. Each region has its own set of collectors, backends, databases, and frontends, and users from each region can access data only from their respective region.
We consider the following two scenarios for federated architecture:
The reason for bringing the time-series data close to the global frontends is lower latency. If latency is high between the regional database and global frontend (> 5 milliseconds), reporting performance on the global frontend could be degraded.
Topology data from regional Collector servers (Topology-Mapping-Service or TMS components) is replicated to the global site Topology-Service, which runs on the global PBE server.
Events data from regional PBE servers is not replicated to the global site but is accessed directly from global frontends.
The ratio of regional to global time-series backend/database instances is 1:1, which means significant additional global resources must be planned beforehand.
Fewer global resources might be used only when no full time-series dataset is replicated from a regional site to the global site. When filters are implemented at regional Data-Gateway (DGW) components, the replicated metrics are limited to a smaller subset. In this case, different architecture is implemented at the global site with fewer backend/database instances. There are global LBCs running on a dedicated Collector server. Those LBCs receive data from regional DGWs and distribute it to target global backends and databases according to the global LBA setup. For a summary description of this architecture, see DR architecture with fewer database resources on replicated site.