The PSM server manages network elements using SNMP and NETCONF. The PSM client and the PSM Dashboard provide you with the user interface to interact with the PSM server.
In the PSM paradigm, network element configuration data resides solely on the network elements themselves. The PSM server does not keep a separate network element configuration database. This absence of a centralized database allows PSM to achieve better scaling and better concurrency.
When management systems operate directly on configuration data stored on the network elements, there is added assurance that the data is accurate and up to date. Because there is only one set of data, issues with synchronizing an external copy of the data with the actual data on the network element do not exist. Furthermore, since the actual data is manipulated directly, it is a simple matter for the management software to detect and handle situations where multiple users attempt to make conflicting changes, regardless of whether the conflicting users are using the CLI, a separate nodal manager, or other PSM servers.
Figure 1 shows a network managed by two PSM servers and multiple clients, along with an external RADIUS server.
When a server comes up, it registers itself as a trap listener on all network elements it is told to discover. It then receives traps from, and reports on, these network elements, and draws topology and discovers services based on configuration and operational data it reads from these network elements. This information is then made available to PSM clients that log in to this server. Clients logging in to a different server might see different network elements altogether, depending on what this other server is configured to discover.