Here's your green button...

dnerdahl

You've done it. At least you've done something. You have jumped through the hoops, danced the dances, shaken the babies and kissed the hands. Now you have a database full of valuable and actionable data and you're presenting it to your customers. Please stop, take a deep breath, and eat a cookie... and possibly some cake. You've earned it. Unfortunately your Sr VP of IT sees you with your hand full of cake and then has THE conversation with you.

S.VP.IT: "I noticed that Solarwinds is collection some really great data. Really well done"

You: *mouth full of cake* "Thanks"

S.VP.IT: "I noticed that there are a lot of things showing as down in the dashboards but there doesn't seem to be anything actually down. There's no open incidents or anything. Is that something you can fix?"

You: "But those are objects that are, in fact, having problems. The best way to clear those up is to get the owners to resolve the issues."

S.VP.IT: "Right. But if there's nothing that's actually down, I'd like for it not to display as down."

You: "But, those objects are actually down."

S.VP.IT: "No no. I get it, but I just want the screen to show green."

That's when it happens. You stuff more cake in your mouth, nod and exit stage right as quickly as possible. You do this because if you don't you'll end up saying something like, "I'll be more than happy to html code you a page in Solarwinds with a big green button on it for you... but it'll be a lie." Executives don't like snarky sarcasm in those moments. If you don't know, trust me, I've had plenty of these conversations when there was no cake.

Brining context to our dashboards for management is an on going issue for Enterprise Systems Monitoring departments and administrators everywhere.

How do you approach the challenge of displaying actionable and accurate data with the need to give management a large sweeping view of their technology with out lying about it?

jrouviere

The myriad of approaches you can take depends on the environment, the tool you're using and the data.  I've had this conversation with customers about patching levels and other things as well, so here's a high level of some of the things I would check and I'll try to throw in a few examples:

1)  Make sure everything's set up correctly and reporting correctly.  If the node isn't actually down, what's the problem?  If the agent service is stopped and triggering the "down" status and you can't get the appropriate team to act, can you temporarily suppress this reporting or if it's an Orion status can you temporarily set it to a simple ICMP status instead of using the agent?

2)  Make sure to resolve the actual issues or remove the pieces that aren't issues.  The example here has more to do with Patch Manager which was the last time I spoke to someone about "make it be 100%"/all green.  The report is as accurate as the data it is given.  If you see services that are down, say Orion based services.  Are they actually used?  If not, uninstall them.  If they are patches that aren't being installed, maybe they aren't applicable and should be declined and/or deleted from WSUS.  I've had all of these happen.  Maybe my mind is on the one product, but when someone asks you why the Orion services are down on miscellaneous server Y, ask them why they're installed there at all.

3)  Sometimes you need to massage the data.  As you state, C-level executives want to see everything be green and 100%.  Often times these guys don't have the technical knowledge to know that isn't possible, or why it is how it is, or maybe they have bigger fish to fry and don't care for the explanation.  Is there another view or filter where you can get the data they want without violating the spirit of reporting in the first place.  In Orion is it something where they have too much access to view unnecessary nodes.  Make sure they see only the correct scope, use account limitations and other methods to make sure they're seeing the key data.  I'm less a fan of this one because in a real world if you understand the data and have a fulfilling conversation with a manager you would need to be creative with how you display the data.  The facts are that some environments are complex, or policy or managerial reasons or what have you make this an option that some people take.  Especially if they're the reporting arm and there's an actual Sysadmin/Security/Networking team responsible for performing the work.

In summary, maybe overly simplified, your tools tend to tell you exactly how it is or at least report on the data you were given:

1)  Make sure the configuration is correct.

2)  Fix anything that's actually a problem.

3)  Make sure you're only revealing as much as is necessary or provide the proper context if possible.