Ron Stone af747f6e15 String cleanup
Replaced some downstream terms with substitutions
Removed double dash (--) in file name and adjusted index
Fixed broken label
Incorporated comment from patchset 1

Signed-off-by: Ron Stone <ronald.stone@windriver.com>
Change-Id: I199481424a9630ad828c4570ac2ff0cbbb4f7c60
Signed-off-by: Ron Stone <ronald.stone@windriver.com>
2021-04-23 11:39:38 -04:00

6.3 KiB

Basic Terminology

The following definitions describe key concepts and terminology that are commonly used in the community and in this documentation.

All-in-one Controller Node

A single physical node that provides a controller function, worker function, and storage function.

Bare Metal

A node running without hypervisors (for example, application workloads run directly on the operating system which runs directly on the hardware).

Worker

A node within a edge cloud that is dedicated to running application workloads. There can be 0 to worker nodes in a edge cloud.

  • Runs virtual switch for realizing virtual networks.
  • Provides L3 routing and NET services.

In a configuration running OpenStack, a worker node is labeled as 'compute' and may be referred to as a compute node.

Controller

A node within a edge cloud that runs the cloud management software (control plane). There can be either one or two controller nodes in a edge cloud.

  • Runs cloud control functions for managing cloud resources.
  • Runs all OpenStack control functions, such as managing images, virtual volumes, virtual network, and virtual machines.
  • Can be part of a two-node HA control node cluster for running control functions either active/active or active/standby.
Data Network(s)

Networks on which the OpenStack / Neutron provider networks are realized and become the VM tenant networks.

Only worker-type and all-in-one-type nodes are required to be connected to the data network(s). These node types require one or more interface(s) on the data network(s).

Deployment Tools

Tools that make the process of distributing, installing, and managing updates.

Edge Computing

The delivery of computing capabilities to the logical extremes of a network in order to improve the performance, operating cost and reliability of applications and services. By shortening the distance between devices and the resources that serve them, and also reducing network hops, edge computing mitigates the latency and bandwidth constraints of today's Internet, ushering in new classes of applications.

From Open Glossary of Edge Computing <https://github.com/State-of-the- Edge/glossary/blob/master/edge-glossary.md#edge-computing>

Infra Network

A deprecated optional network that was historically used for access to the storage cluster.

If this optional network is used, all node types are required to be connected to the INFRA network.

IoT (Internet of Things)

A system of computing devices that can operate with little/no human interaction.

IPMI Network

An optional network on which Intelligent Platform Management Interface (IPMI) interfaces of all nodes are connected. The network must be reachable using L3/IP from the controller's OAM interfaces.

You can optionally connect all node types to the network.

Kubernetes Cluster

A set of machines that has a common control plane for running orchestrated applications.

Management Network

A private network (that is, not connected externally), typically 10GE, used for the following:

  • Internal OpenStack / monitoring and control.
  • VM I/O access to a storage cluster.

All nodes are required to be connected to the management network.

Node

A computer that is usually a server-class system.

Node Interfaces

All nodes' network interfaces can, in general, optionally be either:

  • Untagged single port.
  • Untagged two-port LAG and optionally split between redundant L2 switches running vPC (Virtual Port-Channel), also known as multichassis EtherChannel (MEC).
  • VLAN on either single-port ETH interface or two-port LAG interface.
Network

The network on which all external platform APIs are exposed, (that is, REST APIs, Horizon web server, , and ), typically 1GE.

Only controller type nodes are required to be connected to the OAM network.

Open Source StarlingX

A set of services that can be used to build cloud infrastructure. The source code of the services is available under an open source license that allows access and re-distribution of the codebase. The software components are created and maintained with an open development process through collaboration.

PXEBoot Network

An optional network for controllers to boot/install other nodes over the network.

By default, controllers use the management network for boot/install of other nodes in the OpenStack cloud. If this optional network is used, all node types are required to be connected to the PXEBoot network.

A PXEBoot network is required for a variety of special case situations:

  • Cases where the management network must be IPv6:
    • IPv6 does not support PXEBoot. Therefore, you must configure an IPv4 PXEBoot network.
  • Cases where the management network must be tagged:
    • Most server's BIOS do not support PXEBooting over tagged networks. Therefore, you must configure an untagged PXEBoot network.
  • Cases where a management network must be shared across regions but individual regions' controllers want to only network boot/install nodes of their own region:
    • You must configure separate, per-region PXEBoot networks.

is an open source, complete cloud infrastructure software stack for the edge used by the most demanding applications in industrial , telecom, and other use cases. The platform creates a fusion between OpenStack and Kubernetes to provide a robust and flexible environment for all kinds of workloads, let them be containerized, virtualized or bare metal.

Storage

A node within a edge cloud that is dedicated to providing file and object storage to application workloads. There can be 0 or more storage nodes within a edge cloud.

  • Runs CEPH distributed storage software.
  • Part of an HA multi-node CEPH storage cluster supporting a replication factor of two or three, journal caching, and class tiering.
  • Provides HA persistent storage for images, virtual volumes (that is, block storage), and object storage.
Virtualization

The act of creating a virtual version of CPU, network or storage device.

Virtual Machines (VM)

An instance of a node provided by software (a hypervisor), which runs within the host operating system and hardware.