Nagios Support Forum

I am working on a project with Network Analyzer which will push the limits of what it can do, especially from the point of view of performance. The company I am working with has 200,000 switch/router ports that they want to push through NA. These 200,000 ports will be divided into 4 regions with 4 NA servers. So...I am looking at some ideas to help performance.

Areas of Performance:
These areas are givens that are generally believed to be the most important.

DIsk - looking at solid state drives
Any recommendations on brands that might be the best?

Memory
If NA has 50,000 ports reporting netflow, what are the needs for memory?

CPU
Again, with 50,000 ports what is a guess on CPUs required to work efficiently?

Issues we have been testing:
1. removing atime on disks for performance
2. using RAM disk for at least some aspects
3. rrdcached
Anyone have a working system with rrdcached for NA?
4. restricting logs to only NA, no syslog
5. hdparm and using trim

Any ideas would be helpful...thanks.

Honestly, we have never tested NNA on this size of deployment. As just testing in the environment sounds like it will cost the clients big, I honestly don't know what to tell you.
My main concern is that: even if we could find a solution to collect the amount of data at the rate you wish to do so, due to the amount of data, reporting may have issues working correctly, if at all.