5G Deployment Guide¶

Deployment Overview¶

SD-Core is released with Helm charts and container images. We recommend using Kubernetes and Helm to deploy SD-Core. All SD-Core container images are hosted on Docker Hub and GitHub Registry.

Hardware Resource Requirements¶

CPU & Memory Requirements for 5G components¶
SD-Core Component	Required CPU	Required Memory in Gi
AMF	2 CPU Cores	4Gi
SMF	2 CPU Cores	4Gi
NRF	2 CPU Cores	1Gi
NSSF	2 CPU Cores	1Gi
AUSF	1 CPU Cores	1Gi
PCF	1 CPU Cores	1Gi
AF	1 CPU Cores	1Gi
UDR	2 CPU Cores	1Gi
UDM	2 CPU Cores	1Gi
Config5G(webconsole)	1 CPU Cores	1Gi
SimApp	1 CPU Cores	1Gi
MongoDB	2 CPU Cores	4Gi
UPF-Adapter	1 CPU Cores	1Gi
SCTP Load Balancer	1 CPU Cores	1Gi

Note

SD-Core deployment is tested on Intel/AMD hardware. There is WIP to run SD-Core on ARM architecture.

Deployment Options¶

Development Environments¶

For setting up a 5G development environment, please refer to Aether OnRamp. The legacy Aether In a Box - 4G is preserved for historical reference.

Production Environments - 5G¶

To install SD-Core into your Kubernetes cluster, follow these steps:

Step 1: Clone SD-Core 5G Helm Chart¶

Clone the repository and update dependencies:

git clone https://github.com/omec-project/sdcore-helm-charts.git
cd sdcore-helm-charts/sdcore-helm-charts/
helm dep update  # Update Helm dependencies

Step 2: Prepare Your Helm Values for 5G¶

While you can modify the existing values.yaml directly, we recommend creating a separate values file (e.g., myvalues.yaml) using values.yaml as a template. This approach makes it easier to track your customizations and upgrade in the future.

For detailed explanations of the supported Helm values, refer to the Configuration section below.

Step 3: Install 5G Using SD-Core Umbrella Helm Chart¶

The following command deploys the SD-Core Helm chart with release name sdcore-5g in the sdcore-5g namespace:

helm install -n sdcore-5g --create-namespace -f myvalues.yaml sdcore-5g \
  ./sdcore-helm-charts

Note

Adjust the chart path (./sdcore-helm-charts) to match your local directory structure where you cloned the sdcore-helm-charts repository.

Verify the Installation

Check the Helm release status:

helm -n sdcore-5g ls
xxxx@node:~$ helm -n sdcore-5g ls
NAME        NAMESPACE       REVISION        UPDATED                                 STATUS          CHART           APP VERSION
sdcore-5g   sdcore-5g       1               2022-03-05 16:25:32.338495035 -0700 MST deployed        sd-core-0.10.9
xxxx@node:~$

xxxx@node:~$ kubectl get pods -n aether-5gc
NAME                                           READY   STATUS    RESTARTS   AGE
af-794f6bc999-8z9td                            1/1     Running   0          84m
amf-5d84ccb587-x8b2r                           1/1     Running   0          84m
ausf-57c665bd98-r9t9x                          1/1     Running   0          84m
kafka-dual-0                                   1/1     Running   0          84m
kafka-entity-operator-76764c7dfc-mptj8         2/2     Running   0          83m
metricfunc-6bdb886f6d-jmn8s                    1/1     Running   0          84m
mongodb-0                                      2/2     Running   0          84m
mongodb-1                                      2/2     Running   0          83m
mongodb-arb-0                                  2/2     Running   0          84m
mongodb-kubernetes-operator-7989c65684-bvlxw   1/1     Running   0          84m
nrf-5b76f8cbdc-fb8ws                           1/1     Running   0          84m
nssf-749598c758-lwkzm                          1/1     Running   0          84m
pcf-6c77f966d5-j86p9                           1/1     Running   0          84m
simapp-6549f79d79-rh6dq                        1/1     Running   0          84m
smf-5f7bcc456b-95567                           1/1     Running   0          84m
strimzi-cluster-operator-7c667c9ccb-7xtcn      1/1     Running   0          84m
udm-5bc879dc6c-4dqhf                           1/1     Running   0          84m
udr-59f87695b4-sbs4s                           1/1     Running   0          84m
upf-0                                          5/5     Running   0          84m
webui-76c549747b-6xv2r                         1/1     Running   0          84m
xxxx@node:~$

Application Function (AF)

The 5g-control-plane chart deploys the AF by default (config.af.deploy: true). The AF registers at the NRF, discovers the PCF through it (set pcfUri in config.af.cfgFiles to skip discovery) and exposes a northbound HTTP API on port 7782 (Service af) for the IMS P-CSCF. Set config.af.deploy: false if no IMS is used. The northbound API is unauthenticated unless northbound.authTokenFile is set in afcfg.yaml, so restrict access to it to the IMS network.

Uninstall SD-Core

To uninstall the SD-Core deployment:

helm -n sdcore-5g uninstall sdcore-5g
kubectl delete namespace sdcore-5g  # Optional: remove the namespace

Cloud Native Configuration - 5G¶

Following configuration need to be enabled in 5G helm values override file. It is important to understand usage of following flags

SCTP Load Balancer : Enable this flag to introduce SCTP Load Balancer between gNBs and multiple AMF instances to load balance sctp connections across multiple AMF instances.
DB Store : Enable this flag to preserve AMF context or SMF PDU session into Database. This is required for any AMF/SMF instance to load session/context of any other instance which being fault recovered.
UPF-Adapter : Enable this flag to introduce UPF-Adapter between multiple SMF instances and UPF. This required for the case where UPF does not support multiple SMF association with same pfcp node-id.
NRF Keep-Alive : Enable this flag for NRF to maintain multiple NF profiles and trigger periodic profile updates from the registered NFs.
UE IP-Address alloc via UPF : Enable this config to get UE IP-Address allocated via UPF rather than locally by SMF.
Static UE IP-Address alloc : Enable this config to reserve static UE IP-Address for any specific UE.
Custom IMSI support : Employ this config to have custom IMSI(starts with leading zeroes) for development environment with real UE.