cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post

5 Easy Methods for Improving Your Dashboards

Dashboards are important. Your NOC is an essential avenue for collecting and relaying information about your network, and combined with a finely crafted set of alerts there’s nothing that can get past you. Not only are dashboards effective, but they just look so stinkin’ awesome when done properly. In this post I’m going to focus on my ‘Dashboard Philosophy,’ which is all about efficiency, information, and design. A dashboard should display the most data possible in the space that you have, it should include pertinent information that summarizes your environment, and it should look good doing it. Let’s talk about what the SolarWinds® Orion® Platform brings to the table to help make our dashboards the best they can be.

  1. NOC Views

Using the NOC view feature is a must. These space-saving views allow you to combine multiple sub-views that can be set on a rotation. Creating one is easy: simply add a new summary view, edit it, then enable left navigation and the NOC view feature. Here you can enter an interval for how often the NOC view rotates between individual sub-views. If you aren’t using NOC views, you’re wasting valuable space on your dashboards! Enter NOC mode, full-screen your browser window, and bask in the glory of a massive canvas to display all your fancy metrics and charts. Rob Boss would be proud.

     2. Network Atlas

Admit it, you both love and hate Network Atlas. It’s an incredibly useful tool that requires a bit of extra patience, but the results can be amazing once you get the hang of it. As Henry David Thoreau probably once said, “SolarWinds Network Atlas is but a canvas for your imagination…” or something like that. Check out this amazing example from THWACK® user spzander​:

pastedImage_17.png

Hungry for more? Here is some of my favorite THWACK content for tuning your Network Atlas skills and getting the creative juices flowing:

10 Hidden Gems in Orion Network Atlas

Using Custom Properties to send messages to your NOC using Network Atlas

The “Show us your Network Atlas Maps” thread

     3. PerfStack

With the release of NPM 12.1 came a game-changing new feature… PerfStack. This new charting tool allows you to quickly and easily create attractive charts that contain the data you need while optimizing page space. PerfStack is what makes you, the monitoring professional, shine when an application owner is looking for a way to view monitoring data for their systems. Check out the original release notes for PerfStack here. Since its first iteration, the SolarWinds team has been putting a lot of work into this tool. With PerfStack 2.0, they have added support for many major Orion modules including VMAN, SAM, VNQM, NCM, and DPA, along with a pile of new features such as fast polling, syslog/trap support, quick links, and full screen mode (which makes a great dashboard). As of this post, the next iteration of PerfStack is available in the latest NPM 12.3 Release Candidate and includes… drumroll please… A PERFSTACK WIDGET FOR YOUR DASHBOARDS!

pastedImage_18.png

Here we have a node detail view… WITH PERFSTACK! You can do the same thing with any view type in Orion, including Summary Views (which means dashboards). For dashboard nerds such as myself, this is truly a good day. Sign up for the NPM RC program for more details and awesome sneak peeks at what SolarWinds is doing to improve tools like PerfStack.

     4. AppStack

This is really one of the most efficient ways to display a mass amount of information in such a small space. AppStack is a one-size-fits-all tool that will satisfy your devs, their managers, and your director. An efficient dashboard should have MAXIMUM information in MINIMUM space, and AppStack is the answer. Whether you only have SAM or you’re running multiple products on the Orion platform the AppStack widget gives you a flexible, filterable, and fun-tastic (I couldn’t think of another word that started with ‘f’) resource to add to your dashboards and NOC views. There’s not much more to say. It’s the perfect widget for my Dashboard Philosophy.

pastedImage_19.png

     5. SWQL and Other Advanced Methods

Are you a dev nerd? Do you like to yell at code until it bends to your will? Are you ready to bring your SolarWinds deployment to an unreasonably awesome level? With a little bit of fidgeting and some help from THWACK, you can create your own charts, tables, dashboards, maps, and much more. Check out this post from THWACK MVP CourtesyIT, which has a master list of all the amazing ideas and customizations that have been posted in the community. Be sure to check out the section from THWACK MVP wluther:  he’s got some great content specifically tailored to dashboards. One thing to always keep in mind when using more advanced methods… SolarWinds support may not be able to assist you with the bending of spacetime. Fidget at your own risk!

In my opinion, one of the most powerful tools for creating custom resources is SWQL, the SolarWinds Query Language. With it, data is your slave. THWACK MVP mesverrum makes it easy in this post, where he provides an awesome example of how to create your own custom SWQL tables.

     Results

Let’s put all this together and create a shiny new dashboard that follows the idea of efficiency, information, and design. We need something that doesn’t waste space, contains useful data, and looks awesome. Something like this:

pastedImage_20.png

First thing’s first… we’re using the NOC view, indicated by the black bar at the top with the circles in the upper-right corner that represent the various sub-views in rotation. We have a map from Network Atlas (upper left), a PerfStack project added as a widget (lower left), AppStack (lower right), and a custom SWQL table that displays outage information (check out mesverrum​'s post about it here).

And there we have it! Five useful tools that you can use to make your dashboards amazing. Be sure to post your creations in the community. Here are some threads for NOC views and Network Atlas maps. Now go forth and dashboard!

25 Comments
MVP
MVP

Dashboards are a must have, and nickzourdos​ does a great job at streamlining the essential requirements for a good dashboard.

MVP
MVP

A great read nickzourdos​!  A perfect example of how to combine the various resources to provide a really useful dashboard.

You sir, do an amazing job of writing new custom widgets and resources to make dashboards INCREDIBLE. Major props to you and all the amazing people who have contributed to Eric's master thread of customizations, you folks are the real heros

great read nickzourdos

mikesky​ This is something we can look at for when you are out here this week, especially Atlas and maybe even AppStack.

Level 11

Thanks for sharing!

MVP
MVP

I do like Atlas.

But (this is a 'very big but'):

In fluid environments I've seen too many maps with orphaned nodes (networks with 2000+ nodes).  Add in the logically and physically separate system monitoring those same nodes (redundant SolarWinds systems), and my mapping headaches start to get exponential.  So I've leveraged dynamically populating groups and used that grouping to show what a map would have.  You could say I've thrown the baby out with the bathwater, but with priorities shifting in the blink of an eye, I think maps have been casualties in environments I have been in.  For a cool demo, I would feel OK with some maps.  I can appreciate their creative and visual benefits.  Perhaps the networks I've supported have been more fluid with more disconnect between folks deploying hardware and the monitoring team.  If monitoring is treated as an afterthought, then I've found maps to exacerbate the situation, despite all the cool stuff I could display with them.

MVP
MVP

That's part of what I was trying to address in the post that Nick linked to.  When I was still terribad at scripts and sql all I had was manually building maps.  Once people get some sql under their belts they can get clever and convert tedious manual and error prone tasks like updating maps to a coded solution that automatically adds and removes objects as needed.  These days I mostly rig up a few scripts to generate dynamic groups that do nearly as good a job of summarizing info, but take 1% of the effort on my part and never require further manual maintenance.

Agreed, dynamic groups are the only way to manage maps for large environments. Heck, they're the only way to manage groups for large environments ​period

I vote for more articles like this one.  I'd love to see others' dashboards and learn how to improve my own based on ideas that other SW customers have developed.

Level 10

Is there a limit on number of metrics on one perfstack dashboard?

i can add more number of metrics but what i observed- SWIS is consuming more resources when i try to open the dashboard(~200 metrics) and it's not loading properly.

It helps me a lot  if i get some info on the limitations.

nickzourdos​: Hope you have some info on this.

Level 15

I too found this an interesting and thought-stimulating article.  You identify the essentials and I would find it interesting to see a gallery of what other users have created for dashboards.  I will be the first to admit, I am an engineer, then a scientist, and not much room left for artistry  

Keep articles like this flowing!!!!

Check out the two threads mentioned at the bottom of the post, they are AWESOME!

Great question, I'm not sure there's any documented limitation for Perfstack itself. I'd imagine that eventually you would hit a wall with more than just SWIS, your website is going to have trouble displaying and updating data in a reasonable amount of time.

Level 16

Awesome write up  nickzourdos ! Thank you

Level 12

If only there was a way to animate this.  2018-08-17 09_09_21-Orion Network Atlas.png

Love it!

Level 9

Greaaaaat post!!

I am curious about the perfstack widget you have in the node details view.  What kind of info are you showing and is it dependent on which node you select?

The Perfstack widget is like a map in the sense that it is its own independent 'entity'. You can add any Perfstack project to any dashboard as a widget, whether its relevant to the node/application/etc in that view or not.

MVP
MVP

It's not quite as dynamic as one might hope, but to make a dynamic link to a perfstack chart that changes based on the node you are viewing you can use a variation on the technique that wluther discussed in this post, Building Simple PerfStack Project Lists With SWQL

For example, I use it with ${nodeid} as part of the below Custom Query widget to make it so that when you click on the CPU load it takes you to a dynamic perfstack with just cpu load and alert info on it.  Memory and latency and packet loss also each have their own charts when you click their links.

pastedImage_1.png

select distinct --n.caption as [Node], n.detailsurl as [_Linkfor_Node],

isnull(alerts.[Active Alerts],0) as [Active Alerts]

,CASE

when alerts.[active alerts] is null then '/Orion/images/ActiveAlerts/Check.png'

WHEN sev.Severity = 4 THEN '/Orion/images/ActiveAlerts/Critical.png'

WHEN sev.Severity = 3 THEN '/Orion/images/ActiveAlerts/Serious.png'

WHEN sev.Severity = 2 THEN '/Orion/images/ActiveAlerts/Warning.png'

WHEN sev.Severity = 0 THEN '/Orion/images/ActiveAlerts/InformationalAlert.png'

WHEN sev.Severity = 1 THEN '/Orion/images/ActiveAlerts/Notice.png'

END AS [_iconfor_Active Alerts]

,case when n.cpuload < 0 then 'Not Polled'

when n.host.nodeid is not null and n.host.cpucorecount is not null then concat(round(n.host.cpuload,0),'% of ',n.host.CpuCoreCount,' CPU')

when cpu.[cpu count] is not null then concat(cpuload,'% of ',cpu.[cpu count],' CPU')

else 'Polling Error'

end as [CPU Load]

,'/ui/perfstack/?presetTime=last7Days&charts=0_Orion.Nodes_'+tostring(nodeid)+'-Orion.CPULoad.AvgLoad,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.CPULoad.MaxLoad,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Alerts;' as [_linkfor_CPU Load]

,CASE

WHEN cpuload >= htc.Level2Value THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN cpuload >= htc.Level1Value THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN cpuload < htc.Level1Value  THEN '/Orion/images/StatusIcons/Small-Up.gif'

WHEN cpuload >= n.CpuLoadThreshold.Level2Value and htc.Name is null THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN cpuload >= n.CpuLoadThreshold.Level1Value and htc.Name is null THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN cpuload <  n.CpuLoadThreshold.Level1Value and htc.Name is null THEN '/Orion/images/StatusIcons/Small-Up.gif'

END AS [_IconFor_CPU Load]

,case when percentmemoryused < 0 then 'Not Polled'

else concat(percentmemoryused,'% of ',(round(n.totalmemory/1073741824,0)),' GB')

end as [Memory Used]

,'/ui/perfstack/?presetTime=last7Days&charts=0_Orion.Nodes_'+tostring(nodeid)+'-Orion.CPULoad.AvgPercentMemoryUsed,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.CPULoad.MaxMemoryUsed,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.CPULoad.TotalMemory,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Alerts;' as [_linkfor_Memory Used]

,CASE

WHEN percentmemoryused >= htm.Level2Value and htm.Name = 'VIM.Hosts.Stats.MemUsage' THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN percentmemoryused >= htm.Level1Value and htm.Name = 'VIM.Hosts.Stats.MemUsage' THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN percentmemoryused < htm.Level1Value and htm.Name = 'VIM.Hosts.Stats.MemUsage' THEN '/Orion/images/StatusIcons/Small-Up.gif'

WHEN percentmemoryused >= n.percentmemoryusedThreshold.Level2Value THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN percentmemoryused >= n.percentmemoryusedThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN percentmemoryused <  n.percentmemoryusedThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Up.gif'

END AS [_IconFor_Memory Used]

,CASE

WHEN responsetime<0 then 'No Response'

ELSE concat(responsetime,' ms')

END AS [Latency]

,'/ui/perfstack/?presetTime=last7Days&charts=0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Status,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.ResponseTime.MaxResponseTime,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Alerts,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.ResponseTime.AvgResponseTime;' as [_linkfor_Latency]

,CASE

WHEN responsetime >= n.responsetimeThreshold.Level2Value THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN responsetime >= n.responsetimeThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN responsetime <  n.responsetimeThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Up.gif'

END AS [_IconFor_Latency]

,concat(percentloss,'%') as [Packet Loss]

,'/ui/perfstack/?presetTime=last7Days&charts=0_Orion.Nodes_'+tostring(nodeid)+'-Orion.ResponseTime.Availability,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Status,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.PerfStack.Alerts,0_Orion.Nodes_'+tostring(nodeid)+'-Orion.ResponseTime.PercentLoss;' as [_linkfor_Packet Loss]

,CASE

WHEN percentloss >= n.percentlossThreshold.Level2Value THEN '/Orion/images/StatusIcons/Small-Critical.gif'

WHEN percentloss >= n.percentlossThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Warning.gif'

WHEN percentloss <  n.percentlossThreshold.Level1Value THEN '/Orion/images/StatusIcons/Small-Up.gif'

END AS [_IconFor_Packet Loss]

--,n.percentmemoryusedThreshold.Level1Value

from orion.nodes n

left join (SELECT count(NodeID) as [CPU Count], nodeid

FROM Orion.CPUMultiLoadCurrent

group by nodeid) cpu on cpu.nodeid=n.nodeid

left join Orion.VIM.HostThresholds htc on htc.host.NodeID=n.NodeID and htc.Name = 'VIM.Hosts.Stats.CPULoad'

left join Orion.VIM.HostThresholds htm on htm.host.NodeID=n.NodeID and htm.Name = 'VIM.Hosts.Stats.MemUsage'

left join (SELECT count(AlertActiveID) as [Active Alerts], aa.AlertObjects.RelatedNodeID

FROM Orion.AlertActive aa

group by aa.AlertObjects.RelatedNodeID) alerts on alerts.relatednodeid=n.nodeid

left join (SELECT aa.AlertObjects.RelatedNodeID

,max(CASE

WHEN aa.AlertObjects.AlertConfigurations.Severity = 2 THEN 4

WHEN aa.AlertObjects.AlertConfigurations.Severity = 3 THEN 3

WHEN aa.AlertObjects.AlertConfigurations.Severity = 1 THEN 2

WHEN aa.AlertObjects.AlertConfigurations.Severity = 4 THEN 1

WHEN aa.AlertObjects.AlertConfigurations.Severity = 0 THEN 0

END) AS [Severity]

FROM Orion.AlertActive aa

group by aa.AlertObjects.RelatedNodeID

) sev on sev.relatednodeid=n.nodeid

where nodeid=${nodeid}

order by n.cpuload

Level 9

Interesting.  I'll have to take a closer look at these queries.  THANKS.

Level 12

Very nice read, should have looked for this sooner. Here's to cooler, more informative maps.

Coolness.  I am unworthy to even click on it.

But it's already in my production view for all nodes.  Thanks!

Level 12

Is there an easy way to drop the node name into your query?  I'd like to apply this to a page I have for a group of nodes.  I can do it now, but without the names of the nodes there, the user would be guessing. 

MVP
MVP

If you notice the first line has a chunk commented out with --, remove those dashes and it will drop in the node caption with a clickable link to the details page