URL Monitoring: how are other companies using this?
We are going through the agent-less plug-ins as part of our strategy of checking on the application status from the outside in using monitors.
The URL monitor monitor plug-in uses the hosts as the domain name for the constructed URL; however, in our world, it is poor form to use the host name, because we have a fair bit of 3DNS GTM and F5 BigIP devices that will do URL re-writing and redirecting, so only when the actual alias, and not the underlying host, works.
We're debating extending out this plug-in to provide a way to explicitly support entering domain names, and ignoring the host names if specified. The other option is to have these names specified in the hosts area, but that seems very clunky and has some security aspects as well. Ultimately, we're trying to duplicate the HP SiteScope URL monitoring capability.
We've got a few of these set up. Like you indicated we created a host that is the alias name. I'm not quite sure what security concern there is around this. We also have a bunch of sitescope URL monitors, but quite frankly they were only useful in the most simplest case. The URL transaction monitor in sitescope always seemed too cumbersome and we were able to do stuff faster by writing a script in groovy around httpclient.
Overall we kind of like the control the canoo webtest language that the URL transaction monitor in dynatrace uses and we tend to use this for even simple url checking with dynatrace.
Have you ever considered our syntehtic monitoring offering? http://www.compuware.com/en_us/application-performance-management/products/user-experience-management/synthetic-web-and-mobile/overview.html
It checks availability from "the outside". It also integrates with dT so that PurePaths get automatically tagged from the incoming requests and that you can even drill from the Monitoring Portal to the dynaTrace WebStart Client.
If you want to stick with the agent-less monitor plugins of dynaTrace then I suggest the best option is to define your own hosts in the infrastructure configuration and giving them the actual host names that you want to test
Hi Andi – yes, we've looked at that, but our users are locked behind a bunch of firewalls, so we tend to use internal-based synthetic monitors.
