2 Deployment Environment
This section provides information about the cloud native infrastructure used for Oracle Communications Cloud Native Core, Network Repository Function (NRF) benchmarking.
2.1 Deployed Components
This section provides details about the deployed components.
2.1.1 Hardware Details
This section describes the hardware details.
Table 2-1 Hardware Details
| Nodes | Type |
|---|---|
| Master Nodes | ORACLE SERVER X8-2 |
| Top of Rack Switch | Cisco 93108TC-EX |
| Worker Nodes | ORACLE SERVER X8-2 |
2.1.2 System Software
This section describes the system software details.
Table 2-2 System Software
| System Software | Details |
|---|---|
| Operating System (+Kernel Version) |
5.15.0-105.125.6.2.2.el8uek.x86_64 Oracle Linux Server 8.8 |
| Hypervisor | BareMetal Server |
| CNE | 23.4.1 |
| OSO | NA |
| Kubernetes | 1.27.x |
| ASM | 1.14.6 |
| cnDBTier | 25.1.100 |
2.1.3 Observability Services
This section describes the observability services used for NRF benchmarking.
Table 2-3 Observability Services
| Service Name | Version |
|---|---|
| OpenSearch | 2.3.0 |
| OpenSearch Dashboard | 2.3.0 |
| FluentBit | 1.9.4 |
| Fluentd OpenSearch | 1.16.2 |
| Kyverno | 1.9.0 |
| Grafana | 9.5.3 |
| Prometheus | 2.44.0 |
| Jaeger | 1.45.0 |
2.2 Resource Requirements
The CPU and RAM resources that each service consumes are constrained, so that they do not consume excess resources that could be used by applications. Each service is initially allocated a CPU and RAM at the time of deployment and is allowed to grow to a specified upper limit of each resource while it continues to run. For services where little growth is expected, or where increasing the CPU or RAM underneath a running application might cause an unacceptable service disruption, the initial allocation and upper limit are set to the same value. The resource requests and limits are given below:
Note:
Max replica count of a service must be adjusted as per the target TPS.2.2.1 CNE Common Applications
This section describes the CNE common applications.
Table 2-4 CNE Common Applications
| Service | Pod replica # | CPU/Pod | Memory (Mi) | ||
|---|---|---|---|---|---|
| Requests | Limit | Requests | Limit | ||
| Prometheus | 2 | 2000 | 2000 | 4096 | 4096 |
| Prometheus Node Exporter | 1 per node | 800 | 800 | 512 | 512 |
| Prometheus Operator | 1 | 100 | 200 | 100 | 200 |
| Prometheus AlertManager | 2 | 20 | 20 | 64 | 64 |
| Prometheus Kube State Metrics | 1 | 20 | 20 | 32 | 100 |
| Promxy | 1 | 100 | 100 | 512 | 512 |
| OpenSearch Master | 3 | 1000 | 1000 | 2048 | 2048 |
| OpenSearch Data | 3 | 1000 | 1000 | 16384 | 16384 |
| OpenSearch Client | 3 | 1000 | 1000 | 2048 | 2048 |
| OpenSearch Dashboard | 1 | 100 | 100 | 512 | 512 |
| occne-metrics-server | 1 | 100 | 100 | 200 | 200 |
| occne-alertmanager-snmp-notifier | 1 | 100 | 100 | 128 | 128 |
| Fluentd OpenSearch | 1 per worker node | 500 | 500 | 1024 | 1024 |
| Jaeger Agent | 1 per worker node | 256 | 500 | 128 | 512 |
| Jaeger Collector | 1 | 500 | 1250 | 512 | 1024 |
| Jaeger query | 1 | 256 | 500 | 128 | 512 |
| MetalLB Controller | 1 | 100 | 100 | 100 | 100 |
| MetalLB Speaker | 1 per worker node | 100 | 100 | 100 | 100 |
| LB Controller (vCNE only) | 1 | 10 | 500 | 128 | 1024 |
| Egress Controller (vCNE only) | 1 per worker node | 10 | 1000 | 200 | 500 |
| Bastion Controller | 1 | 10 | 200 | 128 | 256 |
| Kyverno | 3 | 100 | 200 | 256 | 512 |
2.2.2 NRF Services
Table 2-5 NRF Services Resource Requirements
| Service Name | Min Pod Replica Count per NRF |
|---|---|
| Helm test | 1 |
<helm-release-name>-nfregistration |
2 |
<helm-release-name>-nfdiscovery |
2 |
<helm-release-name>-nfsubscription |
2 |
<helm-release-name>-nrfauditor |
2 |
<helm-release-name>-nrfconfiguration |
1 |
<helm-release-name>-nfaccesstoken |
2 |
<helm-release-name>-nrfartisan |
1 |
<helm-release-name>-nrfcachedata |
2 |
<helm-release-name>-ingressgateway |
2 |
<helm-release-name>-egressgateway |
2 |
<helm-release-name>-alternate-route |
2 |
<helm-release-name>-appinfo |
2 |
<helm-release-name>-perfinfo |
2 |
Where, <helm-release-name> is prefixed in each microservice name. For example, if helm-release-name is "ocnrf", then nfregistration microservice name is "ocnrf-nfregistration".
For more information about other resource requirements, see Oracle Communications Cloud Native Core, Network Repository Function Installation, Upgrade, and Fault Recovery Guide.
2.2.3 cnDBTier Services
Table 2-6 cnDBTier Services Resource Requirements
| Service Name | Min Pod Replica Count per cnDBTier |
|---|---|
| MGMT (ndbmgmd) | 2 |
| DB (ndbmtd) | 4 |
| SQL (ndbmysqld) | 2 |
| SQL (ndbappmysqld) | 2 |
| Monitor Service (db-monitor-svc) | 1 |
| Backup Manager Service (db-backup-manager-svc) | 1 |
| Replication Service - Leader | 1 |
| Replication Service - Other | 0 |
Note:
There is a change in the Memory/Pod resource requirement for Monitor Service (db-monitor-svc) and SQL (ndbappmysqld) services.For more information about other resource requirements and recommended parameter values, see Oracle Communications Cloud Native Core, Network Repository Function Installation, Upgrade, and Fault Recovery Guide.