Scalable Distributed Monitoring
InterMapper’s architecture allows a distributed monitoring approach. Although a single monitoring server makes
for simplicity, it may not meet the needs of all networks. For example, the best network design practices could indicate
the need for multiple InterMapper servers to:
- Monitor tens of thousands of devices, without placing a heavy load on a central server
- Distribute the polling so that the traffic says on high-speed local links, not on expensive WAN links
- Monitor equipment behind a client or corporate firewall at a remote site
- Assign local management responsibility to different sites/regions while retaining a central overview of network
health
- Provide resiliency in the face of network outages — even though WAN links are down, remote sites can monitor
their own regions, and resolve problems locally
Remote Servers Allows Distributed Polling
A single InterMapper server can suffice for small installations with hundreds of devices, or a few thousand in a
single location. Although InterMapper is capable of polling many thousands of
devices at a high rate, a network manager might choose to limit the number of devices monitored from a single server,
and thus, the amount of polling traffic, from any one point in the network.
Instead, it makes sense to install a separate InterMapper server as a “polling engine” at strategic
locations on the network. This isolates most of the monitoring traffic on the high-speed links within that portion of the
network. Only summary information goes back to the central InterMapper server over wide area links.
Central and Remote Administration
It’s critical to be able to view and configure all the separate polling engines from the central site.
InterMapper RemoteAccess, the graphical user interface, gives full access to network administrators.
The central site can have top-level maps that show the status of the conditions at the remote sites, and the network
tech can drill down to view the all the sub-maps.
Central Database
Distributing the polling shouldn’t lead to a loss of a central history. InterMapper Database is a
PostgreSQL database that collects historical data — performance statistics, events, etc. — from multiple
InterMapper servers and saves the data into a central database. If the remote server loses its connection to InterMapper
Database, it saves the data locally until the connection is re-established, at which point data collection resumes.
Works through VPNs and Firewalls
InterMapper’s polling typically remains inside a firewall and does not need special permissions to access the
equipment. All connections between the remote InterMapper server and the central office are protected via SSL
encryption.
However, distributing your polling engines to remote sites may require additional protection. Your central InterMapper
server and the RemoteAccess GUI will connect through direct IP connections, VPNs, and firewalls (through a user-selected
port) to your remote InterMapper server without exposing the traffic to the broader Internet.
|