Architecture for monitoring metrics of network management system

1 . A method for monitoring a multi-tenant network management system deployed in a cloud to manage a plurality of groups of datacenters, the network management system comprising a plurality of groups of service instances, the method comprising:

for each respective group of service instances deployed in the cloud to manage a respective datacenter group:

deploying a metrics collection agent within each service instance, in a namespace that contains the services of the service instance, of the group of service instances to collect metrics from services of the service instance and provide the collected metrics to a metric monitoring service instance of the group of service instances, wherein the metrics collection agent communicates with the services using a namespace-local inter-process messaging channel; and

deploying a metrics collection manager within the metric monitoring service instance of the group of service instances, the metrics collection manager for configuring each of the metrics collection agents deployed within the service instances of the group of service instances.

2 . The method of claim 1 , wherein:

each respective datacenter group belongs to a respective tenant; and

each respective group of service instances that manages a respective datacenter group is accessible only by the respective tenant to which the respective datacenter group belongs.

3 . The method of claim 2 , wherein the multi-tenant network management system comprises, in addition to the plurality of groups of service instances, a set of multi-tenant services deployed in the cloud that are accessible by a plurality of the tenants.

4 . The method of claim 3 further comprising:

deploying a particular metrics collection agent to collect metrics from the multi-tenant services and provide the collected metrics to a multi-tenant metric monitoring service; and

deploying a multi-tenant metrics collection manager for configuring the particular metrics collection agent.

5 . The method of claim 4 further comprising deploying a central metrics manager that is accessible by the plurality of tenants to configure metric monitoring for each tenant's respective group of service instances.

6 . The method of claim 3 , wherein the method is performed by one of the multi-tenant services that handles deployment of service instances and the metrics collection agents in the cloud.

7 . The method of claim 1 , wherein:

the network management system is deployed in a Kubernetes cluster in a public cloud; and

for a particular group of service instances, each respective service instance is deployed in a different respective namespace.

8 . The method of claim 7 , wherein deploying a particular respective metrics collection agent within a particular service instance comprises deploying the particular metrics collection agent in the namespace of the particular service instance.

9 . The method of claim 8 , wherein for the particular service instance, the particular metrics collection agent and the services of the service instance share a namespace.

10 . The method of claim 8 , wherein deploying a particular metrics collection manager within a particular metric monitoring service instance comprises deploying the metrics collection manager in the namespace of the metric monitoring service instance.

11 . The method of claim 8 , wherein deploying a particular metrics collection manager for a particular group of service instances comprises deploying the metrics collection manager in a common namespace so that each of the metrics collection agents deployed to the service instances of the particular group of service instances can communicate with the particular metrics collection manager.

12 . The method of claim 7 , wherein:

each of the service instances of a group of service instances is deployed as a set of microservices; and

each microservice is implemented as one or more Pods in the namespace of the service instance to which the microservice belongs.

13 . The method of claim 12 , wherein the metrics collection agent for a particular service instance collects metrics from the Pods implementing at least a subset of the microservices of the particular service instance.

14 . The method of claim 1 , wherein a particular metrics collection manager for a particular group of service instances configures the metrics collection agents deployed within the service instances of the particular group of service instances through a remote procedure call (RPC) channel.

15 . A non-transitory machine-readable medium storing a program which when executed by at least one processing unit deploys entities for monitoring a multi-tenant network management system that is deployed in a cloud to manage a plurality of groups of datacenters, the network management system comprising a plurality of groups of service instances, the program comprising sets of instructions for:

for each respective group of service instances deployed in the cloud to manage a respective datacenter group:

deploying a metrics collection agent within each service instance, in a namespace that contains the services of the service instance, of the group of service instances to collect metrics from services of the service instance and provide the collected metrics to a metric monitoring service instance of the group of service instances, wherein the metrics collection agent communicates with the services using a namespace-local inter-process messaging channel; and

deploying a metrics collection manager within the metric monitoring service instance of the group of service instances, the metrics collection manager for configuring each of the metrics collection agents deployed within the service instances of the group of service instances.

16 . The non-transitory machine-readable medium of claim 15 , wherein:

each respective datacenter group belongs to a respective tenant; and

each respective group of service instances that manages a respective datacenter group is accessible only by the respective tenant to which the respective datacenter group belongs.

17 . The non-transitory machine-readable medium of claim 16 , wherein the multi-tenant network management system comprises, in addition to the plurality of groups of service instances, a set of multi-tenant services deployed in the cloud that are accessible by a plurality of the tenants.

18 . The non-transitory machine-readable medium of claim 17 , wherein the program further comprises sets of instructions for:

deploying a particular metrics collection agent to collect metrics from the multi-tenant services and provide the collected metrics to a multi-tenant metric monitoring service; and

deploying a multi-tenant metrics collection manager for configuring the particular metrics collection agent.

19 . The non-transitory machine-readable medium of claim 18 , wherein the program further comprises a set of instructions for deploying a central metrics manager that is accessible by the plurality of tenants to configure metric monitoring for each tenant's respective group of service instances.

20 . The non-transitory machine-readable medium of claim 15 , wherein:

the network management system is deployed in a Kubernetes cluster in a public cloud;

for a particular group of service instances, each respective service instance is deployed in a different respective namespace; and

the set of instructions for deploying a particular respective metrics collection agent within a particular service instance comprises a set of instructions for deploying the particular metrics collection agent in the namespace of the particular service instance.

21 . The non-transitory machine-readable medium of claim 20 , wherein for the particular service instance, the particular metrics collection agent and the services of the service instance share a namespace.

22 . The non-transitory machine-readable medium of claim 20 , wherein the set of instructions for deploying a particular metrics collection manager within a particular metric monitoring service instance comprises a set of instructions for deploying the metrics collection manager in the namespace of the metric monitoring service instance.

23 . The non-transitory machine-readable medium of claim 20 , wherein the set of instructions for deploying a particular metrics collection manager for a particular group of service instances comprises a set of instructions for deploying the metrics collection manager in a common namespace so that each of the metrics collection agents deployed to the service instances of the particular group of service instances can communicate with the particular metrics collection manager.

24 . The non-transitory machine-readable medium of claim 20 , wherein:

each of the service instances of a group of service instances is deployed as a set of microservices;

each microservice is implemented as one or more Pods in the namespace of the service instance to which the microservice belongs; and

the metrics collection agent for a particular service instance collects metrics from the Pods implementing at least a subset of the microservices of the particular service instance.