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.

Internet up-time

Hello,

   I've been looking into monitoring our Internet Up-time for the purpose of reporting to upper management the metrics of Internet availability within our business. 

There are many reasons why our Internet circuit could go down, and several other reasons in addition that would make our End Users believe the Internet to be down.  I won't list them all here, but the list is made up of typical failures, such as ISP, power, or hardware.  The user experience could be affected if DNS is suddenly unresponsive.

We have NPM, APM, and NTA to help solve this problem, and I would imagine a thorough solution would involve all three. 

While I've only just begun applying the theory to the tools, I was wondering how some of you have thought about this metric.  How do you monitor the User's Internet experience? 

I'm less concerned with what other network related events may be going on concurrently and more interested in knowing when browsing the Internet is just not possible.  (DNS is really an exception to this, as it does rely heavily on the LAN to be functioning.)

Thanks in advance!

 

-JaredC

  • You could setup an APM tempate associated to your internal DNS server.

    Use 1 or more http user experience that goes to a reliable internet source yahoo, google, etc (take out the http://{IP}:{PORT}/ and put http://www.example.com) - that won't actually poll against you DNS server even though it is assigned to it.

    also a dns query for each of those as well (this will poll the DNS server).

    With that you don't care about anything except that a web page is accessible.

  • Also, along with that, if you have multiple internet connections, you can monitor each entrance to the network, by either pinging the default gateway of the different internet carriers. I know this will only provide a low layer work around, but it will show if the DG is replying. 

  • Here is a method I have used that can test via multiple gateways.  Pick something like a public DNS server such as 4.2.2.2 or 8.8.8.8

    Then in your solarwinds server add route statements pointing out your various internet gateways.  Say you have two gateways.  10.0.0.1 and 10.0.0.2

    route add 4.2.2.2 mask 255.255.255.255 10.0.0.1 -p

    route add 8.8.8.8 mask 255.255.255.255 10.0.0.2 -p

    (The -p arguement makes these persistent so they will remain after a reboot)

    This will force the traffic out the path you want so you can monitor a device past your default gateways.  You can use a webpage however the route statements don't work off of a DNS name so you have to specify an IP which can be risky if the IP addresses change.  These servers have been accessible for as long as I can remember.

    You then add the nodes to monitoring and call them something like Verizon GW Test & At&T gateway test.  I find this to be a little more effective than testing just to the carrier gateway because this proves we can get past the carrier. 

    Yet another more complicated option is possible depending on if your gateways have routing knowledge of one another.  Using the same type of route statements we can test one carrier against another.

    Say you have two carriers  Verizon is 63.x.x1  and the internal gateway is 10.0.0.1 and AT&T is 12.x.x.1 and the internal gateway is 10.0.0.2

    You can add two route statements as such

    route add 12.x.x.1 mask 255.255.255.255 10.0.0.1 -p  (this will test the AT&T circuit through the verizon)

    route add 63.x.x1 mask 255.255.255.255 10.0.0.2 -p (this will test the Verizon circuit through the AT&T)

    I prefer this method because it test my ability to go out and come all the way back in, this method will be more complicated if you share a firewall between the two carriers.