MID servers to discover GCP cloud resources and Anthos Kubernetes

[Ashwani Arya1]

Hi,

We have a cloud discovery project to discover Google cloud resources and Anthos Kubernetes clusters. Can some one please advise on the following:

1. What is the best approach to discover Google cloud resources and Anthos Kubernetes clusters? 
2. Can we discover both by placing MID servers on Google cloud or we need to place mid servers for GCP and Anthos Kubernetes separately or do we need to use containerized mid server for Anthos Kubernetes discovery?
3. What are the pros and cons of using VMs in GCP v/s using containerized MID servers?

Thanks,

AA

[Ram Devanathan1]

For any public clouds - to do the cloud resources discovery, you need a mid with the right capabilities and access to internet. Mid can be connected via proxy from your on-prem VMs - it does not have to be running in a cloud VM.

whether you are running your mid on-prem or in cloud, placement of your mids is important - if your mid network is geographically far away, then there's some latency that can creep in.

Note - you are always recommended to be on latest Patterns, CMDB models app version. Please check in store.servicenow.com for latest version.

to Discover google cloud resources, you use the patterns discovery, you can setup a schedule in the Discovery wizard (from left nav Discovery -> Home -> click View active Schedules -> click add cloud schedule). select GCP in the next screen for provider type and then enter all needed information. you can provide your Google project name and discover all related projects in the same org or you can provide your google folder name in the project field, and all projects in the folder and its sub-folders will be listed for discovery - after the validation is done.

you can pick and choose the ones you like or you can select all.

https://youtu.be/GUEbaxo9AeI

apart from the resources you get via regular 'cloud resources' and server-less patterns you have the resource inventory pattern - you can use the resource inventory pattern inclusion list to add more resources not getting discovered via the patterns.

https://docs.servicenow.com/en-US/bundle/tokyo-it-operations-management/page/product/service-mapping...

also look into this -  https://support.servicenow.com/kb?id=kb_article_view&sysparm_article=KB0824021

for kubernetes (GKE) anthos - you have 2 aspects to consider.

one is the GKE k8s cluster inventory itself - this comes for the resources discovery if you turn on the property sn_itom_pattern.k8s_create_schedule_enabled - set it to 'true' (default is 'false'). this will get the list of all gke clusters (as well as other cloud k8s).

the above only gets list of the k8s clusters from cloud and creates the k8s cluster resource records.

then there's the other part which is the details for each cluster.  this would cover on-prem clusters like openshift too.

one way to do this - the detailed cluster components are discovered via kubernetes pattern - this requires the api key connection to the k8s controller. you can use regular mids for this.

https://docs.servicenow.com/en-US/bundle/tokyo-it-operations-management/page/product/service-mapping...

another way to do this - use the cloud native operations app - https://docs.servicenow.com/bundle/sandiego-it-operations-management/page/product/cloud-native-opera...

this deploys the mid containers to run as a micro-service within the cluster - so deploy once to each cluster. this will discover the cluster and give up-to-the-minute updates of latest changes.

hope this helps.

Ram

[Ashwani2]

@Ram Devanathan1 Thank you very much for the detailed response. Can you also please advise on how to calculate number of MID servers required for GCP, Azure cloud resources and Kubernetes components? Is there any calculator or any reference for us to calculate the number of MID servers?

Thanks,

AA