This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

IP Range size for NPM sonar discovery

I am at a bit of a loss as to what size of an IP range to give to NPM for a network discovery job..  I was given the range of addressees that we use in ororganaization and lowest to highest in each address range was from x.x 0.0 to x.x 255.255.  Thats a range of >60k ip adresses!  And is just one of 10 or so I have to run the discovery process for.

What is a manageable range for discovery?   Thats not going to take a year to finish?

We have a single application server which is a virtual machine and its adequately specced with 16GB RAM and 2 cores

Thanks in advance

  • It depends on a lot of factors. You gave us part of one, the server. It also depends on connectivity between the server and all the nodes, what combination of ICMP/WMI/SNMP you're using, how many resources there are to discover on each node, how many nodes actually exist in those ranges, etc.

    I would start small with a Class C and see how it goes, just trial and error. It's likely going to take you a few weeks to finish. There's no reason you can't have overlapping discovery jobs, either. For example define a range that takes roughly 24 hours in your environment, and kick off one at the beginning of the day and one at the end of the day each day.

  • I have about 6 class B address to scan but I've split them all into /24. That way it's quick and easy to see where what is and what is available. I create a folder for each class B address and then when you click on it, it will open the 256 /24 subnets.

    I then have a separate folder that just has each site listed and which ranges are there. This is the folder that always needs to be kept updated, whereas the /24 folders, just update themselves.

  • Things to bear in mind:

    • Maximum run time of the discovery task is 10 hours
    • The responsiveness of your network and parts of it are specific to you
    • On average a class C discovery scan which is 50% populated will take around 10-15 minutes to complete
    • Often when presented with this setup by customers i.e. several Class B networks often they have only a few hundred devices, so you can determine the true population of the networks to determine how you break up these scans

    My advice - Break this up into smaller chunks and do so on two criteria; total number of addresses to scan keeping a single discovery job to no more than 5,000 and logical structure i.e. geographic or business unit

    Mark Roberts

    Prosperon - UK SolarWinds Partners

    Installation | Consultancy | Training | Licenses

    facebook_icon.jpglinkedin.pngblogger.pngtwitter-icon.jpg 

  • All

    I appreciate the input.   Thanks

    T

  • Should it be timing out from a range of 5000 IPs?   C class  range x.x.9.0 thru x.x.30.255 

  • It depends on your settings. Make sure your hop count is set to 0 hops. These are the settings I use for scanning 2 x /16 subnets configured as 1 scan job. Mine doesn't time out.

    Capture.JPG