For me, the presentation layer is always the most important in any monitoring solution. Besides digesting alerts, I’d look for something that gives a ‘big picture’ view (or at least, as big a picture as is relevant for the operations team). If I can’t glance at a screen and quickly identify hotspots, performance issues then it makes the job a lot harder. Building good dashboards also allows me to display meaningful information at the back of the NOC or even just on the IT department’s wall.
SCOM natively gives us the ability to create delegated dashboards that provide limited views into specific resources. From there I can create or customize dashboards to meet the operations teams needs.
My goal here was to create views that draw me to hotspots in my environment at a glance, or give me that ‘all is well’ feeling on the wall of the NOC. If I am displaying these dashboards on a NOC screen there are some considerations:
- you might want to consider a ‘NOC Viewer’ role in SCOM so that you can scope the dashboards and displayed appropriately. This will also discourage people from wanting to poke around any other MPs you may have deployed.
- you might want to consider using a dashboard rotation tool like http://www.dashboardrotator.com. I found this tool doesn’t log me off while it’s rotating dashboards, which is nice (and it’s only $25, which is nicer). If that is an issue then SCOM supports single sign-on too (thanks to @IanNoble for the find.
Below are some sample custom dashboards displayed in web views. These dashboards were built in part with widgets from the Veeam Management Pack for System Center. Bear in mind that natively SCOM requires Internet Explorer and Silverlight to work successfully. See my previous post for some tips on delegated web consoles with SCOM and IE.
Here are the sample dashboards with notes…
This is an example of a state-based dashboard scoped to a single VMware cluster and its resources. This is an example of SCOM delegation (set under the Administration node) and the power of a simple state view. I have purposefully grouped by health state (descending) so resources in a critical condition are always at the top. To create this view I used a Grid layout and simply added State Views to each window.
This is an example of a role-based dashboard. In this example the role is that of a virtualisation storage admin. The central widget gives me capacity planning detail so I can quickly see which datastores are struggling for space and IOPs, now and in the immediate future. To create this dashboard I used a couple of tricks: a grid layout for the dashboard, then added the ‘Column’ layout into the bottom Window. This allowed me to add the Veeam Traffic Light performance widget into each of the windows for the Top 10 datastores for the performance metrics I wanted to view. The central pane in the dashboard uses the Veeam Capacity Planning widget, targeted at the VMware Datastores group and configured to include freeGB and IOPs.
This is an example of a role-based dashboard. In this case it is the role of a virtualisation admin that predominantly looks after all clusters. The heatmap in the center shows memory usage (size) and CPU usage (colour). It quickly tells me who are the memory and CPU hogs. The traffic light views underneath are focused on Cluster pressure, giving me ideas on when I my cluster resources will be stressed (note I could use a capacity planning widget for this too). The final widget shows me how balanced my hosts are in terms of VM load. This dashboard uses a grid layout, with a columns embedded in the bottom Window. The central heat map is a widget imported from the Veeam Management Pack for System Center, scoped to a single vSphere cluster (Production, in this example). The class to group the tiles is ‘VMware Cluster’ and each tile represents a ‘VMware Virtual Machine’. Size is memoryActive and colour is CPUUsed%.
This is a sample heatmap designed to offer an ‘At a glance’ view of the performance health of a specific vSphere cluster (called Flexpod) in this example. The gray VMs are in a powered-off state. The VM names that appear twice in the dashboard are actually cold-standby replicas (replicated using Veeam Backup and Replication). This view was created using a ‘Column’ layout. In all three columns the heatmaps (from the Veeam Management Pack for System Center) were scoped to a single VMware cluster (called Flexpod in my example). The class to group the tiles were either the Flexpod cluster itself or simply by VMware Host, each tile represents a VM or Host (in the first column).
This dashboard uses the Veeam traffic light widget to stack rank VM performance within a specific vSphere cluster (cluster name is Flexpod in this example). The bottom-right widget is the native SCOM performance widget, just to provide a change from the Traffic Light widget. This dashboard was created using a simple Grid view.