VCF Operations 9.1 Real-Time Metrics
- Brock Peterson
- 2 hours ago
- 4 min read
By default VCF Operations collects data on 5-minute intervals, which can be turned down as low as 60-second intervals per adapter instance. VCF Operations 9.1 introduces data collection as low as 2-second intervals, which we are calling Real-Time Metrics, this blog will detail how it works.
VCF Operations 9.1 introduces the VMware Services Platform (VMSP), a Kubernetes Cluster running containerized services. Things like Fleet Lifecycle Management, Operations for Logs, vIDB, Software Depot, Real-Time Metrics, and more. A Simple Mode deployment will include 4 VMs (1 control plane VM and 3 worker node VMs). An HA Mode deployment will have 3 VMs (3 control plane VMs and 3 worker node VMs). Ultimately it will look like this in VCF Operations.
As you can see there are several components, some come with the initial deployment, others are deployed on day-2 (Real-time metrics and the Real-time metrics store components being the latter). Once the Real-time metrics and Real-time metrics store components have been deployed, you'll have access to Real-Time Metrics in Operations. The deployment of the Real-time metrics component is beyond the scope of this blog, but documentation can be found here.
There are three levels of Real-Time Metrics, enabled in Policies.
Essential - roughly 350 metrics spread across Datacenters, Clusters, NSX Transport Nodes, ESXi Hosts, and VMs (20-second data points)
Standard - another ~200 metrics for ESXi Hosts and VMs (20-second data points)
ESX Top - another ~12 metrics for ESXi Hosts (2-second data points)
By default, the Essential Real-Time Metrics are enabled. If you'd like the additional Standard Real-Time Metrics you must enable them in your Policy/s.
If you'd like the additional ESX Top Real-Time Metrics you must also enable them in your Policy/s.
Note: enabling Standard and/or ESX Top Real-Time Metrics in your Policy/s requires a scaling up of your Real-Time Metrics component in Build - Lifecycle - VCF Management - Components - Real-time metrics. By default, the Real-Time Metrics component will be of size Small, to support Standard and/or ESX Top Rreal-Time Metrics it must be scaled up to size Large.
So, where can we find these Real-Time Metrics? Three places:
Operate - Workbench
PromQL Viewer Widget in a Dashboard
Real-Time Metrics API: https://<vcf-instance-services-fqdn>/data-query-service/swagger-ui/index.html
Let's explore these. First, Real-Time Metrics can be found via the Operate - Workbench blade. Here we're looking at a VM, but you could do the same for other Real-Time Metrics supported object types.
Once you've selected an object, the Real Time Analysis tab will show the Real-Time Metrics. The OVERVIEW top left provides an overview of the selected metric class top right. For the case above, we're looking at the CPU-related Real-Time Metrics for this VM. As you hover over each data point you'll see they are being collected every 20 seconds.
You can also use the TOP N tab to show Real-Time Metrics, in this case I'm show the Top 10 VM CPU Readiness % times on the selected ESXi Host.
CUSTOM QUERY is the third available tab and can be used to create your own query (using PromSQL syntax). Here I've created a query for VM Network Received Throughput.
Once built, click RUN on the right and you'll be given that metric as it relates to your selected object.
You can also select a query from the Sample Queries tab and run that.
Click RUN to see the results.
Keep in mind these are all 20-second data points. If you were to select one of the ESX Top provided metrics (indicated by feature='ESX_TOP' in the query), you'd see 2-second data points.
Finally, top right of the OVERVIEW, TOP N, and CUSTOM QUERY tabs, you'll notice 3 dots giving you the option to Save as a PDF, Copy to clipboard, or Add to Notes. The first two can be used for easy sharing, the Add to Notes option can be used to add the chart to the Notes Tab in the Workbench, which can be useful for troubleshooting.
The second place you'll find Real-Time Metrics is in Dashboards using the PromQL Viewer Widget.
I'm driving the PromQL Viewer Widget based on the VM select, it looks like this.
I've chosen a metric from the Metrics tab, but you could also select a query from the Sample Queries tab. Once selected and pasted into the Query field you can confirm it works by clicking the VALIDATE button at right. Once validated, click SAVE buttom right. Documentation on the PromQL Viewer Widget can be found here.
Finally, the third place we can get Real-Time Metrics data is via the API.
There is a handy link provided via Build - Developer Center - APIs & SDKs. Click the Real-time metrics API documentation link which will take you here.
I constructed the URL as indicated, which takes me here.
From here I can do things like run PromQL queries.
This makes Real-Time Metrics available for consumption to tools supporting PromQL extraction techniques.
Real-Time Metrics is a powerful new feature in Ope[rations 9.1, hope this was helpful!
Comments