1
0
mirror of https://github.com/openshift/openshift-docs.git synced 2026-02-07 00:48:01 +01:00
Files
openshift-docs/modules/installation-azure-limits.adoc
2020-12-04 16:40:59 -05:00

118 lines
4.5 KiB
Plaintext

// Module included in the following assemblies:
//
// * installing/installing_azure/installing-azure-account.adoc
// * installing/installing_azure/installing-azure-user-infra.adoc
[id="installation-azure-limits_{context}"]
= Azure account limits
The {product-title} cluster uses a number of Microsoft Azure
components, and the default
link:https://docs.microsoft.com/en-us/azure/azure-subscription-service-limits[Azure subscription and service limits, quotas, and constraints]
affect your ability to install {product-title} clusters.
[IMPORTANT]
====
Default limits vary by offer category types, such as Free Trial and Pay-As-You-Go, and by series, such as Dv2, F, and G. For example, the default for Enterprise Agreement subscriptions is 350 cores.
Check the limits for your subscription type and if necessary, increase quota limits for your account before you install a default
cluster on Azure.
====
The following table summarizes the Azure components whose limits can impact your
ability to install and run {product-title} clusters.
[cols="2a,3a,3a,8a",options="header"]
|===
|Component |Number of components required by default| Default Azure limit |Description
|vCPU
|40
|20 per region
|A default cluster requires 40 vCPUs, so you must increase the account limit.
By default, each cluster creates the following instances:
* One bootstrap machine, which is removed after installation
* Three control plane machines
* Three compute machines
Because the bootstrap machine uses `Standard_D4s_v3` machines, which use 4 vCPUs,
the control plane machines use `Standard_D8s_v3` virtual
machines, which use 8 vCPUs, and the worker machines use `Standard_D4s_v3`
virtual machines, which use 4 vCPUs, a default cluster requires 40 vCPUs.
The bootstrap node VM, which uses 4 vCPUs, is used only during installation.
To deploy more worker nodes, enable autoscaling, deploy large workloads, or use
a different instance type, you must further increase the vCPU limit for your
account to ensure that your cluster can deploy the machines that you require.
By default, the installation program distributes control plane and compute machines across
link:https://azure.microsoft.com/en-us/global-infrastructure/availability-zones/[all availability zones]
within
link:https://azure.microsoft.com/en-us/global-infrastructure/regions[a region].
To ensure high availability for your cluster, select a region with at least
three availability zones. If your region contains fewer than three availability
zones, the installation program places more than one control plane machine in the
available zones.
////
You can [provide an install-config](../overview.md#multiple-invocations) to
[configure](customization.md) the installation program to use specific zones to override the defaults.
////
|VNet
| 1
| 1000 per region
| Each default cluster requires one Virtual Network (VNet), which contains two
subnets.
|Network interfaces
|6
|65,536 per region
|Each default cluster requires six network interfaces. If you create more
machines or your deployed workloads create load balancers, your cluster uses
more network interfaces.
|Network security groups
|2
|5000
| Each default cluster
Each cluster creates network security groups for each subnet in the VNet.
The default cluster creates network
security groups for the control plane and for the compute node subnets:
[horizontal]
`controlplane`:: Allows the control plane machines to be reached on port 6443
from anywhere
`node`:: Allows worker nodes to be reached from the Internet on ports 80 and 443
|Network load balancers
| 3
| 1000 per region
|Each cluster creates the following
link:https://docs.microsoft.com/en-us/azure/load-balancer/load-balancer-overview[load balancers]:
[horizontal]
`default`:: Public IP address that load balances requests to ports 80 and 443 across worker machines
`internal`:: Private IP address that load balances requests to ports 6443 and 22623 across control plane machines
`external`:: Public IP address that load balances requests to port 6443 across control plane machines
If your applications create more Kubernetes `LoadBalancer` service objects,
your cluster uses more load balancers.
|Public IP addresses
|3
|
|Each of the two public load balancers uses a public IP address. The bootstrap
machine also uses a public IP address so that you can SSH into the
machine to troubleshoot issues during installation. The IP address for the
bootstrap node is used only during installation.
|Private IP addresses
|7
|
|The internal load balancer, each of the three control plane machines, and each
of the three worker machines each use a private IP address.
|===