OpenStack is the most advanced virtualization framework based on python programming language. It comprises of all three ‘eyes’ of virtualization – compute, storage and network. The purpose of this blog is to list the various services of OpenStack.

OpenStack implementation is in the form of REST API providers. Each component of it are implemented as services serving REST APIs via GET, PUT, POST and DELETE verbs. Communication between services of an OpenStack component is based on message queue services – RabbitMQ is the popular one. And components talk to each other via REST api. Each component provide its own REST api client for this purpose. For example, if one of the compute service wants to communicate with networking service, nova uses neutron client.

Before we venture into a good list of such services, it will be a good idea to see where the various services are implemented and deployed. While it is possible to put entire OpenStack solution for your favorite hypervisor into one monolithic package and install it on one operating system instance, it will serve for demonstration purposes only. A true and scalable implementation in an enterprise environment requires divide and conquer strategy.

For example, certain openstack services can run on a operating system instance on or around hypervisor (hosting virtual machines). And other services can run on another operating system instance. The later can be called management node and former can be called compute node. Nova – the compute component of OpenStack employs such a division – API and scheduling and orchestration services can be made to run on management node, while its core compute service can run on the compute node. To make this clear, when a virtual machine deployment is needed, certain API of nova api service is invoked on management node to set the ball rolling. Then scheduler and orchestration layers on management node validate and delegate actual virtual machine creation, association of storage boot volumes and network interfaces to their counterpart on compute node.

Cinder – the storage component of OpenStack works with the nova component, to create and attach boot volumes to deploying virtual machine. This is accomplished via a communication between nova api, cinder services on management node with nova compute service on compute node.

Neutron – the network component of OpenStack work similar to cinder services. A neutron api service on management node manages creation of network and sub-network required for network connectivity for virtual machines. It interacts with nova compute service on compute node to attach required network interface with the deploying virtual machine. There is an optional but often required ‘agent’ service running on compute node as well, and it interacts with neutron api service on management node. The purpose of this agent architecture is to ‘bind’ a neutron network/subnetwork/port with the deploying virtual machine. A successful binding of a neutron port with nova instance helps in managing network interface of a virtual machine for additional post-deploy use cases. (like migration or evacuation of a virtual machine to another hypervisor).

 

The following is the list of OpenStack services made available on management node:

1. Keystone – provides identity service used for authorization (by user service) and authentication (by token) purposes.
2. Glance registration and api – for image management.
3. nova api, conductor, scheduler, vnc and console authorization – compute management related services.
4. neutron api, agent, dhcp, l3 and metadata – network management related services.
5. placement api – used for scheduling and filtering of hypervisor data used by nova compute services.
6. cinder api, scheduler and volume – storage services.
7. ceilometer acentral and anotification – monitoring services.

The following is the list of OpenStack services on compute node:

1. Placement api.
2. neutron agent – network agent on compute host.
3. nova api metadata and compute – compute services.
4. ceilometer acentral and anotification – monitoring services.