cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Kubernetes Metrics based Method

Do we have place where we can able get the Kubernetes metrics based on specific method, Like 

Classic Fullstack, Cloud Native, Kubernetes observability, Application only and Host Monitoring. 

and how the HOSTUNITS licenses will be consumed.

7 REPLIES 7

Mizső
DynaMight Guru
DynaMight Guru

Hi @VenkataSainath 

These ara the built in Kubernetes metrics:

Built-in Metrics on Grail - Dynatrace Docs

Regarding hostunits in classic licencing 1 host unit / 16 GB at the nodes in ClassicFullStack; CloudNativeFullStack and Host monitoring. In case of Application only the instrumented pod memory limit is the base of the host unit calculation (eg. 4gb limit 0,25 HU), if there is not pod memory limit the base is the underlaying host entire memory.

I do not understand exactly the Kubernetes observability item in you list, may be it is the Kubernetes API monitoring. It consumes DDU not Host unit in the classic licensing model.

I hope it helps.

Best regards,

Mizső

 

Dynatrace Community RockStar 2024, Certified Dynatrace Professional

Hi @Mizső ,

 

Basically, we are looking the table with all methods with metrics  because in application monitoring method will not get the node level metrics and in Host monitoring will not get container level metrics.  so need know at least in which method which metrics will provide?

Regarding Kubernetes observability , i see the in the deployment page as one of the deployment method.


gopher
Pro

@VenkataSainath  , I believe what you're looking for is here :  Built-in classic metrics - Dynatrace Docs
If you have the 'Kubernetes' monitoring via the Active Gate you will get these.

Basic summary of how it works: 

K8S integration via Active Gate - Basic Node (CPU / Mem), Cluster, workload / pods / containers.  (CAdvisor / Kubestate API)   
Fullstack - Agent on Node + Infra Metrics + Applications
Application Only - Pod / Container code level monitoring via agent injection (trace data + other telemetry)

Hope this helps.

@gopher , Thanks for your info. 

but when we trying to do the testing on HOSTMONITORING method, we are not seeing the containers level metrics.

 

Regards,

Venkat

Hi @VenkataSainath   Host Monitoring will not capture pod or container level metrics. It will only collect physical Node level metrics.

You need the 'Kubernetes Observability'  (AKA - Container Active Gate) deployment to get the required metrics. 
This will scrape all the metrics you need from KubeState & cAdvisor  endpoints,   giving you Cluster, Node, Pod & Container metrics. 

Unless you need to monitor the underlying node (generally not needed if you're on a cloud provider like EKS / GKE / AKS) then there is not that much additional benefit and a potentially significant cost impact of implementing the host monitoring on a Kubernetes cluster.   

If you are not using a containerised active gate, you can enable the Kubernetes API calls to run from a remote active gate , which can be used to capture this data as well.

Hope this helps.

Hi @gopher ,

 

Thanks for your explanation, Do we have any place for this in Dynatrace documentation?

And How Kubernetes Observability deployment consume the Hostunit License 

@VenkataSainath  deployment methods are probably something you should discuss with a Dynatrace rep if you can first. 

But there are typically 2 approaches: 
1. Full stack  (cloud native or classic)
-  you pay for everything in the 'host monitoring' license doc.
-  you pay for everything to be monitored 'by default' on the cluster
This gives you:  
Host monitoring - node monitoring
Application monitoring - container code level insights
Kubernetes monitoring - k8s insights. 

2. Application Monitoring (host) + Kubernetes monitoring 
-   you pay for Kubernetes license and the Application only in the 'host monitoring' license doc. 
-  minimise expenditure by only covering what you need. 
- Use deployment labels to make code level monitoring via application only 'opt-in' 
This gives you:  
Application monitoring - container code level insights
Kubernetes monitoring - k8s insights. 

For deployment of the Kubernetes Monitoring refer to:   
Quickstart - Dynatrace Docs
Get started with Kubernetes observability - Dynatrace Docs  <- this one here

Licensing is here for Kubernetes: Container monitoring - Dynatrace Docs
you would then take the  application only 'container injection' : Host monitoring - Dynatrace Docs   

As mentioned, probably something to speak to Dynatrace about to get the best understanding.
Enjoy.

 

Featured Posts