See Browser Compatibility for additional details.
CloudCenter Suite supports Kubernetes 1.15.4 and earlier releases.
The CloudCenter Suite requires Tiller v2.12.3 to be installed. Refer to the Helm documentation for additional details.
Installers are already incorporated in the CloudCenter Suite SaaS offer, see SaaS Access for additional details.
Resource Requirements for CloudCenter Suite Modules
The following table lists the minimum resource requirements assuming that you install all available modules.
|Public Cloud5||Private Cloud3|
|vCPU||Memory (GB)||Storage (GB)||vCPU||Memory (GB)||Storage (GB)|
Workload Manager4 and Cost Optimizer
Kubernetes Cluster (3 primary servers)
1 Update only one module at a time. If you simultaneously update more than one module, your update process may fail due to limited resource availability.
2 Before updating any module, verify that you have un-allocated CPU/Memory in your cluster to ensure that your environment has free CPU/Memory – a module-update scenario requires additional resources for the old pod to continue running until the new pod initializes and takes over. This additional resource requirement is temporary and only required while a module update is in Progress. After the module is updated, the additional resources are no longer needed.
3 On private clouds (vSphere and OpenStack), each of the 3 primary server instances require 3 vCPU and 8 GB memory and 40 GB storage (root disk), hence the difference in the additional requirement of 9 vCPU, 24 GB memory, and 120 GB storage (root disk). See the Number of VMs section below for additional details. Similarly, each worker instances require 3 vCPU and 8 GB memory and 40 GB storage (root disk) – however, the number of workers changes dynamically at install time. Installer VMs require a minimum of 4 vCPUs and 8 GB RAM.
4 Workload Manager numbers include considerations for 4 Cloud Regions in the same instance. To support additional cloud regions, you must scale your cluster by adding Kubernetes worker nodes. You will need 1 CPU and 3 GB memory for each additional region.
5 Public clouds do not support auto-scaling – the number of nodes might differ if scaled on an auto-scaling enabled node group.
6 The storage is 230 GB just to enable StatefulSet migration. In reality, only 115 GB is being used for operation of services.
Number of VMs
A CloudCenter Suite installation launches a highly available Kubernetes cluster which consists of primary server(s) and worker(s) instances.
The number of worker nodes (for both private and public cloud) vary based on the instance type selected during the installation process.
For private clouds, a redundant cluster requires a minimum of 2 out of 3 primary server nodes to be running at any point, so the cluster can function as designed.
If you plan to scale up at a later date, be aware that the worker instance type selected at installation time will also be used for the scaled nodes.
The CloudCenter Suite requires that the underlying disks for Kubernetes disk attachments be redundant and available. Most public clouds already provide built-in redundancy for their block disks (AWS EBS, GCP Persistent Disks, and so forth). Be sure to verify that the Datastores/Datastore Clusters are also on redundant, non-local storage (NFS, NetApp) before you begin the installation process.
IP Pool Requirements
You must select IP address to ensure that each IP endpoints is available, accessible, and not used by any other resource.
When configuring or modifying you pool of IP addresses, be aware of the following requirements:
Verify if the IP pool can accommodate additional workloads.
Select your instance type according to the following dependencies – based on your instance type selection, the installer displays the error or success information in the UI.
The CloudCenter Suite setup requires 3 primary servers.
The CloudCenter Suite dynamically calculates the number of application VMs (workers).
Do not use 172.18.0.1/16 for the installer instance as this IP address is used by the Docker/Kubernetes setup.
NodePort: If you set the type field to NodePort, the Kubernetes control plane allocates a port from a range specified by – service-node-port-range flag (default: 30000-32767). Refer to https://kubernetes.io/docs/concepts/services-networking/service/ for additional details.
You must either set the Network Time Protocol (NTP) time at the datacenter level or at the time of installation.
If set at installation time, then verify that the network can access the NTP server.
The time for all worker and primary server nodes is synced with the primary controller node. The primary controller node is the instance used to launch the CloudCenter Suite – identified by the link that takes you to the Suite Admin UI (Take Me to Suite Admin). This link contains the IP address of the primary controller node as displayed in the following screenshot.
The Suite Installer Dashboard
After launching the installer, navigate to the IP address of your VM in a supported browser. This presents the Suite Installer Dashboard. The Suite Installer Dashboard has the following options:
Without Internet Access
The Cisco Repository is used to host Cisco-related files and packages for various purposes. You may need to install the CloudCenter Suite in an environment that does not have internet access. If so, you need to set up the offline repository. See Offline Repository to sync your offline repository with the Cisco repository.
As you will be shutting down the installer VM after the installation, you can reuse that VM to set up the offline repository.
- No labels