Sizing for Standalone Installations
The following are suggested sizing guidelines for your Redcell Synergy system.
1 Assumptions: Servers have at least four cores and are no more than four years old. As memory and usage increases, the number of CPU cores increase. Two cores can work for the most basic installations, but are not recommended.
2 Each device is equivalent to a L2 or L3 switch with a total of 48 interfaces per device being monitored. For each of devices not being monitored for 48 interfaces, one can add another 50 devices to the overall inventory for ICMP-only monitoring.
3 Application Constraints are most relevent to Traffic Flow Analysis, Peformance Management, and Event Management.
Traffic Flow Analysis ratings map to constant throughput divided by sample rate, as in bandwidth / sample rate. 20G / 2000 is easier to manage than 20G / 1000. 20G / 1 is a thousand times more demanding than 20G / 1000. Best practice is to avoid such high sample rates. The bandwidth the hardware your Redcell Synergy installation can support is dramatically lower in such cases. Best practice is to sample a maximum of one traffic flow for every 1000 (1:1000). Higher sampling rates degrade database performance and increase network traffic without adding any significant statistical information.
Performance Management can support 600 inserts per second using a single disk (SSD) Drive. 1 insert = 1 monitored attribute. Expect better performance as you add more drives (and worse performance with slower drives).
Event Management can support a sustained 1200 traps /sec using a single (SSD) drive. Expect better performance as you add more drives (and worse performance with slower drives).
4 Database memory settings increase as the number of database hits increases. At the 32GB level best practice is to use an SSD drive or fast disk array because of the large number of database actions possible.