Recently a customer of mine asked for a simpler way to monitor the performance of his vSAN disk groups. He wanted to be able to select a host cluster and view each of the disk groups within that cluster. This can easily be accomplished with vRealize Operations!
This custom dashboard is comprised of four vROps widgets. An object list, a heat map, a scoreboard, and object relationship. The user selects a host cluster from the object list and all of the vSAN disk groups for that cluster are represented on the heat map. Each of the tiles is sized according to the total capacity of each disk group and colored based on the average write latency in milliseconds. This will help guide the user toward any hot spots within their environment. Next, the user selects a disk group from the heat map and all the relevant performance metrics are displayed on the scoreboard and the health of the capacity disks are displayed in the object relationship widget. It’s literally as easy as one, two, three!
If you’ve read my blog post on creating custom dashboards with vROps, then you’ll know how easy this is. In fact, the most difficult part is deciding what metrics to display, but I’m going to show you how we can simplify that also!
Laying out the dashboard
The first thing we need to do is to create a new dashboard. This requires the advanced or enterprise editions of vRealize Operations Manager. In the dashboards view, select actions and click on create dashboard.
Give your dashboard snazzy name. I also like to give a description of how the dashboard is intended to work.
Now, let’s add some happy little widgets! Expand the widget list dropdown and drag an object list, heat map, scoreboard, and object relationship onto the dashboard canvas.
Next, we’ll edit the object list so that it just shows host clusters. Click on the pencil icon located on the object list’s title bar.
Give the object list a title to be displayed on the dashboard and filter for object types -> cluster compute resource.
Click the show columns button on the bottom left corner of the widget.
Uncheck everything except for Name. This will configure the object list to only show cluster names.
Next, we need to configure the heatmap. To do so, click on the pencil icon on the heatmap’s title bar.
The heatmap has the most amount of settings that we need to configure, but even this is easy to do. First, we need to give it a title because calling it heatmap doesn’t tell us anything about what it does.
When nothing is selected in our cluster object list, the heatmap will show all disk groups. This is why we group the blocks by cluster compute resource. Click on the group by drop down and select group by vCenter adapter -> cluster compute resource.
We only want the heatmap to show vSAN disk groups, so select object type -> vSAN adapter -> vSAN disk group.
Next, we need to decide how large to make the blocks on our heatmap. We could leave this alone so that all blocks are the same size, but I like to size them based on the overall capacity of the disk group. Click on the size by drop down and select disk space -> total capacity.
This next one is really up to you. What metric should you use to drive the color of the heatmap blocks? Well, this depends on your workload profiles and vSAN configuration. I.E. are you workloads typically write intensive and are you using a caching tier? In this example, I’ll use average write latency. If you’re really unsure, you could use average device latency and modify this later when you have a better idea of what your key performance indicators (KPI) are. Click on the color by dropdown and select disk I/O -> average write latency (ms).
Okay, we’re done here now. Right? Maybe. This next step is optional, but I recommend you do it. If you don’t set a color range then vROps will automatically scale this for you. The good side is that this is one less decision you need to make. The downside, however, is that this makes the heatmap less readable because your highest value will always be red regardless of whether or not it’s really an issue. For example, let’s say you have four disk groups and all their write latencies are in the microseconds. You have three disk groups that are around 300µs and one that’s 400µs. The 400µs disk group will show as red on the heatmap even though 400µs is well within acceptable operating spec. Avoiding “the sky is falling” scenarios is why I prefer to enter my own values. I normally recommend starting relatively low and adjust accordingly. Here I entered 0ms is green and 150ms is red.
Moving on to the scoreboard! Click on the pencil to edit the widget.
The scoreboard has a few settings that we’ll want to adjust. Give it a title, show three columns, show the object name, metric name, metric unit, and the last 24 hours of sparklines. Next, we could pick each of the metrics that we want to display individually, but someone was nice enough to package the metrics we need in a metric configuration right out of the box!
Let’s talk about Metric Configurations!
Simply stated, metric configurations are a collection of metrics that can be used in things like metric charts and scoreboards. These are really handy if you have a group of metrics that you want to use with multiple dashboards and save you from having to pick or remember individual metrics. Metric configurations are XML formatted and can now be edited directly from within vROps. One of the biggest benefits of these groups being wrapped in a cozy XML blanket is that these are easy to share between multiple vROps environments.
Even if you’re not fluent in XML, vROps has a really handy menu-driven wizard to add metrics to your collection without having to touch the keyboard!
Getting back to the scoreboard configuration, pick the disk group performance metric configuration and click save.
The last widget we need to configure is the object relationship. Click the pencil in the widget’s title bar.
All we need to do here is give the widget a title and filter for adapter types -> vSAN adapter.
Next, we need to expand the widget interactions on the left-hand menu. Our “select cluster” widget is a self-provider which means it’s pulling a list of host clusters by itself and therefore we don’t need to select any providers for this widget. The “select disk group” heatmap, however, will show disk groups based on thbe cluster selected in our object list, so make “select cluster” a provider for our heatmap. Step three is our scoreboard which is displaying the metrics for the disk group selected in the heatmap, so make “2.) select disk group” a provider for the scoreboard. The same goes for the object relationship widget. Make “select disk group” a provider for “capacity disk health.”
Click apply interactions and then click save to complete your dashboard.
Hopefully, this dashboard will be useful to you as well and be sure to check out our next post on implementing views within out dashboards to identify anomalies in your environment faster.