Preparing to install IBM Edge Computing for servers
Before you install IBM Cloud Private, enable IBM Multicloud Manager and configure IBM Edge Computing for servers, and ensure that your system meets the following requirements. These requirements identify the minimum required components and configurations of your planned edge servers.
These requirements also identify the minimum configuration settings for the IBM Multicloud Manager hub cluster that you plan to use to manage your edge servers.
Use this information to help you plan the resource requirements for your edge computing topology and your overall IBM Cloud Private and IBM Multicloud Manager setup.
- Hardware requirements
- Supported IaaS
- Supported environments
- Required ports
- Cluster sizing considerations
Hardware requirements
When you are sizing your management node for your edge computing topology, use the IBM Cloud Private sizing guidelines for a single or multi-node deployment to help you size your cluster. For more information, see Sizing your IBM Cloud Private cluster .
The following edge server requirements only apply to IBM Cloud Private instances that are deployed to remote operations centers by using the edge computing profile.
Requirement | Nodes (boot, master, management) | Worker nodes |
---|---|---|
Number of hosts | 1 | 1 |
Cores | 4 or more | 4 or more |
CPU | >= 2.4 GHz | >= 2.4 GHz |
RAM | 8 GB or more | 8 GB or more |
Free disk space for installation | 150 GB or more |
Note: 150 GB of storage enables up to three days of log and event data retention if there is network disconnection to the central data center.
Supported IaaS
The following table identifies the supported Infrastructure as a Service (IaaS) that you can use for your edge services.
IaaS | Versions |
---|---|
Nutanix NX-3000 Series for use in edge server locations | NX-3155G-G6 |
IBM Hyperconverged Systems that are powered by Nutanix for use in edge servers | CS821 and CS822 |
For more information, see IBM Hyperconverged Systems powered by Nutanix PDF .
Supported environments
The following tables identify the additional Nutanix configured systems that you can use with your edge servers.
LOE site type | Node type | Cluster size | Cores per node (total) | Logical processors per node (total) | Memory (GB) per node (total) | Cache disk size per disk group (GB) | Cache disk quantity per node | Cache disk size per node (GB) | Storage total cluster pool size (all flash) (TB) |
---|---|---|---|---|---|---|---|---|---|
Small | NX-3155G-G6 | 3-node | 24 (72) | 48 (144) | 256 (768) | N/A | N/A | N/A | 8 TB |
Medium | NX-3155G-G6 | 3-node | 24 (72) | 48 (144) | 512 (1,536) | N/A | N/A | N/A | 45 TB |
Large | NX-3155G-G6 | 4-node | 24 (96) | 48 (192) | 512 (2,048) | N/A | N/A | N/A | 60 TB |
LOE site type | Node type | Cluster size | Cores per node (total) | Logical processors per node (total) | Memory (GB) per node (total) | Cache disk size per disk group (GB) | Cache disk quantity per node | Cache disk size per node (GB) | Storage total cluster pool size (all flash) (TB) |
---|---|---|---|---|---|---|---|---|---|
Small | CS821 (2 socket, 1U) | 3-node | 20 (60) | 80 (240) | 256 (768) | N/A | N/A | N/A | 8 TB |
Medium | CS822 (2 socket, 2U) | 3-node | 22 (66) | 88 (264) | 512 (1,536) | N/A | N/A | N/A | 45 TB |
Large | CS822 (2 socket, 2U) | 4-node | 22 (88) | 88 (352) | 512 (2,048) | N/A | N/A | N/A | 60 TB |
For more information, see IBM Hyperconverged Systems that are powered by Nutanix .
Required ports
If you are planning to deploy a remote edge server with a standard cluster configuration, the port requirements for the nodes are the same as the port requirements for deploying IBM Cloud Private. For more information about these requirements, see Required ports . For the required ports for the hub cluster, see the section Required ports for IBM Multicloud Manager.
If you plan to configure your edge servers by using the edge computing profile, enable the following ports:
Port | Protocol | Requirement |
---|---|---|
NA | IPv4 | Calico with IP-in-IP (calico_ipip_mode: Always) |
179 | TCP | Always for Calico (network_type:calico) |
500 | TCP and UDP | IPSec (ipsec.enabled: true, calico_ipip_mode: Always) |
2380 | TCP | Always if etcd is enabled |
4001 | TCP | Always if etcd is enabled |
4500 | UDP | IPSec (ipsec.enabled: true) |
9091 | TCP | Calico (network_type:calico) |
9099 | TCP | Calico (network_type:calico) |
10248:10252 | TCP | Always for Kubernetes |
30000:32767 | TCP and UDP | Always for Kubernetes |
Note: Ports 30000:32767 have external access. These ports must be opened only if you set Kubernetes Service type to NodePort.
Cluster sizing considerations
For IBM Edge Computing for servers, the hub cluster is typically a standard IBM Cloud Private hosted environment. You can use this environment to also host other compute workloads that you need, or want, to be served from a central location. The hub cluster environment must be sized so that it has sufficient resources to host the IBM Multicloud Manager hub cluster and any additional workloads that you plan to host on the environment. For more information about sizing a standard IBM Cloud Private hosted environment, see Sizing your IBM Cloud Private cluster .
If needed, you can operate a remote edge server within a resource-constrained environment. If you need to operate an edge server within a resource-constrained environment, consider using the edge computing profile. This profile configures only the minimum required components that are needed for an edge server environment. If you use this profile, you must still allocate sufficient resources for the set of components that are required for an IBM Edge Computing for servers architecture and for providing the resources that are needed by any other application workloads that are hosted on your edge server environments. For more information about the IBM Edge Computing for servers architecture, see IBM Edge Computing for servers.
While the edge computing profile configurations can save memory and storage resources, the configurations result in a low level of resilience. An edge server that is based on this profile can operate disconnected from the central data center where your hub cluster is located. This disconnected operation can usually last for up to three days. If the edge server fails, the server ceases to provide operational support for your remote operations center.
The edge computing profile configurations are also limited to supporting only the following technology and processes:
- Linux 64-bit platforms
- Non-high availability (HA) deployment topology
- Addition and removal of worker nodes as day-2 operations
- CLI access to, and control of, the cluster
- Calico networks
If you require more resilience or if any of the preceding limitations are too constraining, you can instead select to use one of the other standard deployment configuration profiles for IBM Cloud Private that provide for greater failover support.
Sample deployments
- Edge server environment based on the edge computing profile (resilience low)
Node type | Number of nodes | CPU | Memory (GB) | Disk (GB) |
---|---|---|---|---|
Boot | 1 | 1 | 2 | 8 |
Master | 1 | 2 | 4 | 16 |
Management | 1 | 1 | 2 | 8 |
Worker | 1 | 4 | 8 | 32 |
-
Edge server environments based on other IBM Cloud Private profiles (resilience medium to high)
Use the small, medium, and large sample deployment requirements when you need to use a configuration other than the edge computing profile for your edge server environment. For more information, see Sizing your IBM Cloud Private cluster sample deployments .