I was most recently told to use the SNMP Statistic Polling index number since that takes UnDP's into consideration as well, so in this case your number would be 3405. However, it does make sense to me that you would add it to the Status Index number.
It would be nice to get clarification on that. Ultimately none of it is 100% accurate as your "mileage may vary" depending on what modules you are running, polling frequency, etc.
Also, there was recently a thread talking about how keeping the primary Orion application server lightly loaded will improve the functionality of your entire Orion infrastructure; I have personally found this to be true as well.
Thanks for your insight byrona. I would like someone else to chime in with their .02 to clarify if you could.
The amount of UnDP Pollers do not influence the amount of Elements being Listed.to get a number of how manu UnDPs you have you can run the Following SQL Query: Select Count(*) from CustomPollerAssignment
Elements is a Total of Nodes + Interfaces + Volumes. This is the number we look at when suggesting that you add an Additional Poller.
Pollers which you can see under Settings> Database Details is the amount of Total SNMP Queries the NetPerfMponService is running on the Poller.
Remember when we recommend 8000 - 10000 Elements, this is based upon Default Polling Intervals of 10 Minute Nodes, 9 Minute Interfaces, and 15 Minutes for Volumes.
I am goign to have to get a bit upset about this as I was very specifically told the following by two different SolarWinds employees...
"When determining the need for a different poller using the 8k - 10k limit you need to look at the SNMP Statistic Polling Index number as this takes the UnDP's into consideration as well"
This is not the first time I have received inconsistent information from different people at SolarWinds. My company expects me to be able to track the capacity of our system and I have recently quoted them numbers using the information that was provided by the product vendor only to find out that those numbers are incorrect.
This leave me with two questions...
- What do I need to go in order to get reliable information from SolarWinds regarding the product?
- Why aren't UnDP's taken into consideration consideirng that heavy use of them would significantly alter the load on the system?
Please understand that my frustrations are not specifically with you but more generally about the lack of consistent information that I receive from SolarWinds empolyees.
Sean, I know that you are answering the question but for absolute clarity, please let me know if I am understanding correctly.
When I look at the Database, I see that I have:
6200 Network Elements.
(1804 Nodes, 2007 Interfaces, 2389 Volumes)
I have 9432 Pollers. I have yet to implement APM but will be doing so next week. Are you saying that I am currently below the 8,000-10,000 Elements, regardless of the number of pollers I actually have on the system?
I was under the impression that the total number of Pollers was what you were using to determine the need for another Polling engine.
If I have 1 Node with 2 Interfaces, 5 Volumes, and 17 APM Pollers, are you saying that I can have 8000 of those and be fine?
Please clarify at your earliest convenience as this is a pressing issue regarding our current setup.
SolarWinds: We would really like to get a definitive answer on this. Thus far the answers I have received from SolarWinds on this have been contradictory.
I would like to see a definitive answer from a Product Manager and even better would be to see answered in some official documentation so it is no longer questioned.
Hi Byron et al.
I've marked this for the PM to review.
Thanks for your feedback.