Skip to content

Commit 43b9942

Browse files
committed
Add glossary terms for DRA concepts
1 parent 9eaa956 commit 43b9942

File tree

7 files changed

+166
-0
lines changed

7 files changed

+166
-0
lines changed
Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
---
2+
title: Common Expression Language
3+
id: cel
4+
date: 2025-06-04
5+
full_link: https://cel.dev
6+
short_description: >
7+
An expression language that's designed to be safe for executing user code.
8+
tags:
9+
- extension
10+
- fundamental
11+
aka:
12+
- CEL
13+
---
14+
A general-purpose expression language that's designed to be fast, portable, and
15+
safe to execute.
16+
17+
<!--more-->
18+
19+
In Kubernetes, CEL can be used to run queries and perform fine-grained
20+
filtering. For example, you can use CEL expressions with
21+
[dynamic admission control](/docs/reference/access-authn-authz/extensible-admission-controllers/)
22+
to filter for specific fields in requests, and with
23+
[dynamic resource allocation (DRA)](/docs/concepts/scheduling-eviction/dynamic-resource-allocation)
24+
to select resources based on specific attributes.
Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,23 @@
1+
---
2+
title: Device
3+
id: device
4+
date: 2025-05-13
5+
short_description: >
6+
Any resource that's directly or indirectly attached your cluster's nodes, like
7+
GPUs or circuit boards.
8+
9+
tags:
10+
- extension
11+
- fundamental
12+
---
13+
One or more
14+
{{< glossary_tooltip text="infrastructure resources" term_id="infrastructure-resource" >}}
15+
that are directly or indirectly attached to your
16+
{{< glossary_tooltip text="nodes" term_id="node" >}}.
17+
18+
<!--more-->
19+
20+
Devices might be commercial products like GPUs, or custom hardware like
21+
[ASIC boards](https://en.wikipedia.org/wiki/Application-specific_integrated_circuit).
22+
Attached devices usually require device drivers that let Kubernetes
23+
{{< glossary_tooltip text="Pods" term_id="pod" >}} access the devices.
Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,23 @@
1+
---
2+
title: DeviceClass
3+
id: deviceclass
4+
date: 2025-05-26
5+
full_link: /docs/concepts/scheduling-eviction/dynamic-resource-allocation/#deviceclass
6+
short_description: >
7+
A category of devices in the cluster. Users can claim specific
8+
devices in a DeviceClass.
9+
tags:
10+
- extension
11+
---
12+
A category of {{< glossary_tooltip text="devices" term_id="device" >}} in the
13+
cluster that can be used with dynamic resource allocation (DRA).
14+
15+
<!--more-->
16+
17+
Administrators or device owners use DeviceClasses to define a set of devices
18+
that can be claimed and used in workloads. Devices are claimed by creating
19+
{{< glossary_tooltip text="ResourceClaims" term_id="resourceclaim" >}}
20+
that filter for specific device parameters in a DeviceClass.
21+
22+
For more information, see
23+
[Dynamic Resource Allocation](/docs/concepts/scheduling-eviction/dynamic-resource-allocation/#deviceclass)
Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,25 @@
1+
---
2+
title: Dynamic Resource Allocation
3+
id: dra
4+
date: 2025-05-13
5+
full_link: /docs/concepts/scheduling-eviction/dynamic-resource-allocation/
6+
short_description: >
7+
A Kubernetes feature for requesting and sharing resources, like hardware
8+
accelerators, among Pods.
9+
10+
aka:
11+
- DRA
12+
tags:
13+
- extension
14+
---
15+
A Kubernetes feature that lets you request and share resources among Pods.
16+
These resources are often attached
17+
{{< glossary_tooltip text="devices" term_id="device" >}} like hardware
18+
accelerators.
19+
20+
<!--more-->
21+
22+
With DRA, device drivers and cluster admins define device _classes_ that are
23+
available to _claim_ in workloads. Kubernetes allocates matching devices to
24+
specific claims and places the corresponding Pods on nodes that can access the
25+
allocated devices.
Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,23 @@
1+
---
2+
title: ResourceClaim
3+
id: resourceclaim
4+
date: 2025-05-26
5+
full_link: /docs/concepts/scheduling-eviction/dynamic-resource-allocation/#resourceclaims-templates
6+
short_description: >
7+
Describes the resources that a workload needs, such as devices. ResourceClaims
8+
can request devices from DeviceClasses.
9+
10+
tags:
11+
- workload
12+
---
13+
Describes the resources that a workload needs, such as
14+
{{< glossary_tooltip text="devices" term_id="device" >}}. ResourceClaims are
15+
used in
16+
[dynamic resource allocation (DRA)](/docs/concepts/scheduling-eviction/dynamic-resource-allocation/)
17+
to provide Pods with access to a specific resource.
18+
19+
<!--more-->
20+
21+
ResourceClaims can be created by workload operators or generated by Kubernetes
22+
based on a
23+
{{< glossary_tooltip text="ResourceClaimTemplate" term_id="resourceclaimtemplate" >}}.
Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
---
2+
title: ResourceClaimTemplate
3+
id: resourceclaimtemplate
4+
date: 2025-05-26
5+
full_link: /docs/concepts/scheduling-eviction/dynamic-resource-allocation/#resourceclaims-templates
6+
short_description: >
7+
Defines a template for Kubernetes to create ResourceClaims. Used to provide
8+
per-Pod access to separate, similar resources.
9+
10+
tags:
11+
- workload
12+
---
13+
Defines a template that Kubernetes uses to create
14+
{{< glossary_tooltip text="ResourceClaims" term_id="resourceclaim" >}}.
15+
ResourceClaimTemplates are used in
16+
[dynamic resource allocation (DRA)](/docs/concepts/scheduling-eviction/dynamic-resource-allocation/)
17+
to provide _per-Pod access to separate, similar resources_.
18+
19+
<!--more-->
20+
21+
When a ResourceClaimTemplate is referenced in a workload specification,
22+
Kubernetes automatically creates ResourceClaim objects based on the template.
23+
Each ResourceClaim is bound to a specific Pod. When the Pod terminates,
24+
Kubernetes deletes the corresponding ResourceClaim.
Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,24 @@
1+
---
2+
title: ResourceSlice
3+
id: resourceslice
4+
date: 2025-05-26
5+
full_link: /docs/reference/kubernetes-api/workload-resources/resource-slice-v1beta1/
6+
short_description: >
7+
Represents one or more infrastructure resources, like devices, in a pool of
8+
similar resources.
9+
10+
tags:
11+
- workload
12+
---
13+
Represents one or more infrastructure resources, such as
14+
{{< glossary_tooltip text="devices" term_id="device" >}}, that are attached to
15+
nodes. Drivers create and manage ResourceSlices in the cluster. ResourceSlices
16+
are used for
17+
[dynamic resource allocation (DRA)](/docs/concepts/scheduling-eviction/dynamic-resource-allocation/).
18+
19+
<!--more-->
20+
21+
When a {{< glossary_tooltip text="ResourceClaim" term_id="resourceclaim" >}} is
22+
created, Kubernetes uses ResourceSlices to find nodes that have access to
23+
resources that can satisfy the claim. Kubernetes allocates resources to the
24+
ResourceClaim and schedules the Pod onto a node that can access the resources.

0 commit comments

Comments
 (0)