api-ref: Further clarify placement aggregates

Following up on the comments at [1], provide further disambiguation and
clarity on the difference between placement aggregates and nova host
aggregates, with some illustrative examples that include where the
former may be used to represent the latter (as is being proposed via
[2]).

[1] https://review.openstack.org/#/c/539033/1/placement-api-ref/source/aggregates.inc@14
[2] https://review.openstack.org/#/q/bp/placement-mirror-host-aggregates

Change-Id: I53588fca311992f885c4157591232a864b6a86b1
This commit is contained in:
Eric Fried 2018-02-16 08:52:30 -06:00
parent 0cf7a8a648
commit e8406e6437

View File

@ -7,14 +7,19 @@ resource providers in groups called aggregates. API calls in this
section are used to list and update the aggregates that are associated
with one resource provider.
Provider aggregates are used for modeling relationships among providers.
Examples may include:
* A shared storage pool providing DISK_GB resources to compute node providers
that provide VCPU and MEMORY_MB resources.
* Affinity/anti-affinity relationships such as physical location, power failure
domains, or other reliability/availability constructs.
* Groupings of compute host providers *corresponding to* Nova host aggregates
or availability zones.
.. note:: Placement aggregates are *not* the same as Nova host
aggregates and should not be considered equivalent.
Provider aggregates are used for modeling relationships where a provider
shares resources with one or more other providers. An example is a shared
storage pool providing DISK_GB resources to other compute node providers that
provide VCPU and MEMORY_MB resources.
The primary differences between Nova's host aggregates and placement
aggregates are the following:
@ -37,8 +42,9 @@ aggregates are the following:
providers are individually decorated with *traits* that provide
qualitative information about the resource provider.
* In Nova, a host aggregate dictates the *availability zone* within
which one or more nova-compute services reside. Placement aggregates
have no concept of an availability zone.
which one or more nova-compute services reside. While placement aggregates
may be used to *model* availability zones, they have no inherent concept
thereof.
.. note:: Aggregates API requests are availiable starting from version 1.1.