We are implementing a core and failover unit at our aggregation point which will only receive information from remote probes out in our school districts. Each school will have a single NetFlow sensor and anywhere from 50 to 2500 sensors (mainly on the smaller end) being monitored by their probe.
Question#1: How many sensors would our core be limited to for this scenario?
Question#2: When purchasing hardware for our core server(s), should we lean towards more RAM or more processing power or a balance of both?
Question#3: Is there any advantage for performance with any specific RAID level on the core server?
Hello,
1. Usually the total limit for one single PRTG instance is around 20,000 sensors, where these are mostly "resource - friendly" sensors, as PING and SNMP Sensors.
2. A balance of good CPU and RAM power would be our recommendation.
3. Disk-I/O is not as important as CPU & RAM resources for PRTG, so there is no need to establish an explicit RAID only for PRTG.
best regards.
Apr, 2011 - Permalink