Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

My management asks for a monitoring screen to be displayed inside a hallway. They think about those fancy screenshots salesmen show when they try to sell this wonderfull appliance which never breaks down (never, promise). Or the kind of thing we see when visiting Nagio or Cacti website (I shown some, they said yes).

My management knows nearly nothing about what computing people are doing. They want this to have some professional feelings about network, servers, etc. And probably to be seen as professionnal themselves.

We have :

  • 3 servers (1 fileserver, 2 applications servers)
  • 1 backup server
  • 1 LAN with 3 managed switches
  • 2 ADSL lines
  • 1 VPN connecting 7 small remote sites
  • 1 Oracle database
  • 2 applications running on the servers
  • and lots of small things

We already use Nagios. They are not interested with the simple view we have, saying everything is ok. I'd like to have some better ideas to present to them, so they can tell me they want this or that.

I think about :

  • a pie chart displaying disk usage (with department names)
  • a graph à la Cacti to show disk I/O on the file server
  • one graph showing network throughput on the switches
  • one graph showing Oracle throughput, or something related
  • a weathermap for the VPN

Any better ideas ?

share

locked by HopelessN00b Feb 19 at 7:31

This question exists because it has historical significance, but it is not considered a good, on-topic question for this site, so please do not use it as evidence that you can ask similar questions here. This question and its answers are frozen and cannot be changed. More info: help center.

closed as primarily opinion-based by HopelessN00b Feb 19 at 7:30

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

2  
Would management be amused by a running manhour total and cost calculation? Disk I/O usage: 7%, Cost of displaying technical information to management: $4,000... –  HopelessN00b Dec 24 '12 at 2:42
    
@HopelessN00b : the Hollywood perspective on IT - priceless... @ GregoryMOUSSAT: Anything flashy and constantly changing will do, methinks. –  Deer Hunter Dec 24 '12 at 2:56
    
@Deer Hunter : I also think flashy and moving things will be a must for this kind of "need". I just have no good idea. –  Gregory MOUSSAT Dec 24 '12 at 16:06
    
@GregoryMOUSSAT : BTW, will access to the hallway be restricted? Since you're doing this to secure funding, the politically correct way would be thinking in terms of the bottom line - not dollars/euros but rather number of transactions. Every minute your servers furnish and process information with some latency; they also store new information which creates risk mitigated by your backup policies. Thus, three items: transactions/requests served, latency distribution, and level of information loss risk outstanding would make the composite view understandable. My $.02 ... –  Deer Hunter Dec 24 '12 at 18:46

2 Answers 2

up vote 4 down vote accepted

Interesting problem. I've had to come up with similar dashboards before, but never had to do it for anything but technical management. The sort of display you're looking for... may need custom code to present information from whatever monitoring system you're using. A few of the systems (I'm thinking of Zenoss right now) have enough customizability to maybe build something internally, but for true polish you'll be re-presenting data you're already gathering.

Those sorts of top-level managers are looking for big-picture with the possibility of drilling down. Your displayed items should be the topmost critical items. Without knowing your applications, I'd probably put the following on a dashboard:

  • Network throughput for the Internet connection (graph or dial)
  • A chart showing loading on the Oracle database. I realize this is a complex thing to discover, but find some proxy and display it.
  • An overall disk-space display of some kind. If it changes often enough a graph, or a simple thermometer if it doesn't.
    • If they're concerned about it, charts for each org-unit they care about; can be harder to get.
    • If you're using shared storage, a chart showing unallocated space on your disk array(s) graphed over time. These are big-budget items so are worthy of tracking.
  • Application-level status: is the app up, working in a degraded but functional state, or down?
  • App-server loading.

I'd actually steer clear of the VPN heat-map, unless they really want that. It's useful for pretty information density, which can be a goal all by itself, but I don't think it communicates meaningful information to their level.

Likewise switch throughput, unless backplane bandwidth is something you're actually worried about. If they want the pretty-information, go for it. But if they're wanting useful I'd only add it if there was space left on the display.

Disk I/O may be a good idea for political reasons (backups are really expensive in I/O, golly don't we need more widgets for that), or pretty density, but again not something I'd present to upper management.

share

Interesting question (sorry I'm a bit late to the party here, but hey that's the joy of ServerFault and the internet - this is now here for posterity.)

I think the right answer is to present a combination of "show" and "go" to management. Even if they're non-technical, having some substance and logic behind what you present in a dashboard will make a good impression. You can take the data and present it in any number of visually-appealing ways (charts, graphs, etc.).

I just wrote a non-vendor-specific piece about the subject of monitoring in general called Zen and Art of System Monitoring. Take a peek for some specifics on what to monitor, regardless of the system/application.

In short, the way to think about what's useful (and non-technical management will appreciate this) is to break your systems down into layers and come up with one or two metrics for each layer -- a few key application metrics, process stats, server stats, network stats, and so forth. Which stats are relevant depends on what your applications are doing (are they web servers? network heavy? disk heavy?).

Consider also monitoring not only "bad things" but "lack of good things" -- there's old lore about systems monitoring at Google that said that by monitoring "AdWords $ / sec", you'll get a pretty good view of the whole stack. So figure out what your version of that is an include it in your dashboards.

Hope this is helpful food for thought for any future folks stumbling on this answer.

share

Not the answer you're looking for? Browse other questions tagged or ask your own question.