Implementing an application manifest in a node-specific manner using an intent-based orchestrator

1. A method comprising:

providing a first orchestrator executing in a network computing environment including a plurality of computing nodes, the first orchestrator programmed to deploy a plurality of role instances of a bundled application on the plurality of computing nodes according to an application manifest;

providing a second orchestrator executing in the network computing environment, the second orchestrator programmed with a capacity to select among nodes of the plurality of computing nodes and instantiate containers and maintain availability of containers on the selected nodes; and

for each role instance of the plurality of role instances:

selecting, by the first orchestrator, for the each role instance, a host node for the each role instance;

invoking, by the first orchestrator, the second orchestrator to instantiate a container on the host node executing the each role instance such that the capacity of the second orchestrator to select among the plurality of computing nodes is suppressed;

instantiating, by the second orchestrator, the each role instance and a container executing the each role instance on the host node;

generating, by the first orchestrator, a network plan associating network identifiers with each role instance of the plurality of role instances; and

for each role instance of the plurality of role instances:

invoking, by the second orchestrator, an interface to the first orchestrator;

providing, by the first orchestrator to the second orchestrator, a selected network identifier of the network identifiers associated with the each role instance in the network plan; and

assigning, by the second orchestrator, the selected network identifier to the each role instance.

2. A method comprising:

providing a first orchestrator executing in a network computing environment including a plurality of computing nodes, the first orchestrator programmed to deploy a plurality of role instances of a bundled application on the plurality of computing nodes according to an application manifest;

providing a second orchestrator executing in the network computing environment, the second orchestrator programmed with a capacity to select among nodes of the plurality of computing nodes and instantiate containers and maintain availability of containers on the selected nodes; and

for each role instance of the plurality of role instances:

selecting, by the first orchestrator, for the each role instance, a host node for the each role instance;

invoking, by the first orchestrator, the second orchestrator to instantiate a container on the host node executing the each role instance such that the capacity of the second orchestrator to select among the plurality of computing nodes is suppressed;

instantiating, by the second orchestrator, the each role instance and a container executing the each role instance on the host node; and

for each role instance of the plurality of role instances:

invoking, by the first orchestrator, mounting of a storage volume to the host node for the each role instance prior to invoking the second orchestrator to instantiate the container on the host node executing the each role instance; and

configuring, by the second orchestrator, the container on the host node to access the storage volume.

3. The method of claim 1 , further comprising:

generating, by the first orchestrator, a compute plan specifying host nodes for all role instances of the plurality of role instances.

4. The method of claim 3 , wherein generating the compute plan comprises assigning host nodes for all role instances of the plurality of role instances in accordance with affinity and anti-affinity constraints in the application manifest.

5. The method of claim 4 , further comprising:

detecting, by the first orchestrator, failure of a first node of the plurality of computing nodes hosting a first container executing a first role instance of the plurality of role instances; and

in response to detecting the failure of the first node:

identifying, by the first orchestrator, a second node of the plurality of computing nodes satisfying the affinity and anti-affinity constraints; and

invoking, by the first orchestrator, the second orchestrator to instantiate the first role instance in a second container on the second node.

6. The method of claim 5 , further comprising: instructing, by the first orchestrator, the second orchestrator to assign a same identifier to the second container as was assigned to the first container.

7. The method of claim 5 , wherein detecting failure of the first node comprises:

detecting, by the first orchestrator, that an attempt by the second orchestrator to re-instantiate the first role instance on the first node was unsuccessful.

8. The method of claim 5 , further comprising:

writing, by the first container, changes to a state of the first container to a persistent remote storage volume;

invoking, by the second container, an interface to the first orchestrator; and

instructing the second container, by the first orchestrator, through the interface, to configure its state according to the persistent remote storage volume.

9. The method of claim 1 , wherein the second orchestrator comprises a KUBERNETES master and KUBELETs executing on the plurality of computing nodes.

10. A system comprising:

a network computing environment including a plurality of computing nodes;

a first orchestrator executing in the network computing environment, the first orchestrator programmed to deploy a plurality of role instances of a bundled application on the plurality of computing nodes according to an application manifest; and

a second orchestrator executing in the network computing environment, the second orchestrator programmed with a capacity to select among nodes of the plurality of computing nodes and instantiate containers and maintain availability of containers on the selected nodes;

wherein the first orchestrator and the second orchestrator are programmed to, for each role instance of the plurality of role instances:

select, by the first orchestrator, for the each role instance, a host node for the each role instance;

invoke, by the first orchestrator, the second orchestrator to instantiate a container on the host node executing the each role instance such that the capacity of the second orchestrator to select among the plurality of computing nodes is suppressed; and

instantiate, by the second orchestrator, the each role instance and a container executing the each role instance on the host node; and

wherein the first orchestrator and the second orchestrator are programmed to:

for each role instance of the plurality of role instances:

invoke, by the first orchestrator, mounting of a storage volume to the host node for the each role instance prior to invoking the second orchestrator to instantiate the container on the host node executing the each role instance; and

configure, by the second orchestrator, the container on the host node to access the storage volume.

11. A system comprising:

a network computing environment including a plurality of computing nodes;

a first orchestrator executing in the network computing environment, the first orchestrator programmed to deploy a plurality of role instances of a bundled application on the plurality of computing nodes according to an application manifest; and

a second orchestrator executing in the network computing environment, the second orchestrator programmed to select nodes of the plurality of computing nodes and instantiate containers and maintain availability of containers on the selected nodes;

wherein the first orchestrator and the second orchestrator are programmed to, for each role instance of the plurality of role instances:

select, by the first orchestrator, for the each role instance, a host node for the each role instance;

invoke, by the first orchestrator, the second orchestrator to instantiate a container on the host node executing the each role instance such that the capacity of the second orchestrator to select among the plurality of computing nodes is suppressed; and

instantiate, by the second orchestrator, the each role instance and a container executing the each role instance on the host node; and

wherein the first orchestrator and the second orchestrator are programmed to:

generate, by the first orchestrator, a network plan associating network identifiers with each role instance of the plurality of role instances; and

for each role instance of the plurality of role instances:

invoke, by the second orchestrator, an interface to the first orchestrator;

provide, by the first orchestrator to the second orchestrator, a selected network identifier of the network identifiers associated with the each role instance in the network plan; and

assign, by the second orchestrator, the selected network identifier to the each role instance.

12. The system of claim 10 , wherein the first orchestrator is further programmed to generate a compute plan specifying host nodes for all role instances of the plurality of role instances.

13. The system of claim 12 , herein the first orchestrator is further programmed to generate the compute plan by assigning host nodes for all role instances of the plurality of role instances in accordance with affinity and anti-affinity constraints in the application manifest.

14. The system of claim 13 , wherein the first orchestrator and the second orchestrator are programmed to:

detect, by the first orchestrator, failure of a first node of the plurality of computing nodes hosting a first container executing a first role instance of the plurality of role instances; and

in response to detecting the failure of the first node:

identify, by the first orchestrator, a second node of the plurality of computing nodes satisfying the affinity and anti-affinity constraints; and

invoke, by the first orchestrator, the second orchestrator to instantiate the first role instance in a second container on the second node.

15. The system of claim 14 , wherein the first orchestrator is further programmed to instruct the second orchestrator to assign a same identifier to the second container as was assigned to the first container.

16. The system of claim 14 , wherein the first orchestrator is programmed to detect failure of the first node by:

detecting that an attempt by the second orchestrator to re-instantiate the first role instance on the first node was unsuccessful.

17. The system of claim 14 , wherein:

the first container is programmed to write changes to a state of the first container to a persistent remote storage volume;

the second container is programmed to invoke an interface to the first orchestrator; and

the first orchestrator is programmed to instruct the second container, through the interface, to configure its state according to the persistent remote storage volume.

18. The system of claim 10 , wherein the second orchestrator comprises a KUBERNETES master and KUBELETs executing on the plurality of computing nodes.