How can you monitor a website external to the companies network when the site is hosted internally?
In other words, we've had an issue here where a website will work internally but not externally.
Any ideas?
you either need a routing path that APM can take to come out a different path than that of your internal users.
or
a APM instance hosted in a different site
SEUM
with a playback location external to your internal users as above.
Option 1: Routing will be an issue as the network here is controlled abroad and the new security lockdowns on the network won't allow it.
Option 2: Very interested to know more. We have an external broadband line here onsite that is not part of our network. Can we utilize this in anyway for this APM?
Option 3: You lost me at SEUM.
thanks
also.. if you have a machine externally you could install a Proxy on it, which effectively routes the web traffic via that server.
could you provide a high level explanation of this scenario?
I don't have much of a networking background... sorry
Here's a diagram of how I montor the performance of our external facing web server.
I use 3 external nodes to get network diversity (network outages outside our network are out of our control) and by having 3 different testing nodes, I can tell if the "Internet" is broken or is it our problem.
Option b requires a second install of APM and thus more licenses. Many enterprise users will place a APM instance regionally and then interconnect them with the EOC product.
You may be able to dual home your host with the dsl line and route out that way using option A but that could get tricky with host routing on your setup.
Option C is a new SW product for transactional monitoring of web sites but it would still require a box that sits on the outside.
Crgegors suggestion is similar in scope to SEUM but homebrewed and quite inventive. I was headed down a similar path when SEUM was released.
Do you know what the root causes of your last few outages have been?
You describe the outside not working but internal users have access.
If its something like DNS not in sync internal vs external it could be an easy issue to address with other monitors.
I saw the preview of SEUM but will wait untill it handles painfull things like Oracle Forms based apps.
For Simple (web 1.0 type) web apps I've found that using a linux box and the nagios check_http monitor works just fine.
For the setup I describe above, each of the external Linux nodes have the nagios monitors installed locally. I then created Linux/Unix Script Monitor using the template that SolarWinds provides for nagios integration.
Then I call the script with the command line:
${SCRIPT} /usr/lib64/nagios/plugins/check_http -c 25 -4 -H www.example.com
This forces the check_http plugin to use IPV4 and set a critical error time at 25 seconds. 25 is pretty high, but I use APM to apply thresholds to the response times I get back from the plugin.
Chris.
SEUM is really the best solution. I wrote a blog posting tackling this exact issue a few weeks ago.
