Hardware requirements
Learn about the hardware requirements for an IBM Cloud Pak® for Watson AIOps deployment on Red Hat® OpenShift® Container Platform.
You can choose from two sizes for your IBM Cloud Pak for Watson AIOps deployment: small or large, depending on your requirements. Information is provided in the following tables on the event throughput and the number of resources that can be monitored for each deployment size, and the minimum hardware requirements for each size.
Sizing for IBM Cloud Pak foundational services is included. For more information, see Hardware requirements and recommendations for foundational services 
.
Notes:
- You cannot change your selected size after deployment.
- The resources per compute node (worker nodes) in the following table are an example of an optimal configuration. If different compute node configurations are used, it can result in oversizing CPU and memory resource.
- You require extra worker nodes if you install the optional Advanced Agile Discovery extension. The Advanced Agile Discovery extension is composed of the application discovery and network discovery services. If you have a large sized deployment, you require an extra worker node for each of the Advanced Agile Discovery service extensions that you install. If you have a small deployment, you require one extra worker node if you have one or both of the Advanced Agile Discovery service extensions installed. For more information about the Advanced Agile Discovery extension, see Installing extensions.
- Multi-region and multi-zone clusters are not supported.
Supported resource number and throughput rates
The following tables detail the number of resources that can be monitored and the maximum event rate throughput for the IBM Cloud Pak for Watson AIOps components for each of the deployment sizes.
Event management component
| Category | Resource | Small | Large |
|---|---|---|---|
| Event rate throughput | Steady state events per second Burst rate events per second |
20 100 |
50 500 |
Topology Management component
| Category | Resource | Small | Large |
|---|---|---|---|
| System size | Approximate resources | 200,000 | 5,000,000 |
| Environment options | Container size | Trial | Production |
| Environment options | High availability | No | Yes |
Hardware requirements for OpenShift and IBM Cloud Pak for Watson AIOps
The following table shows the total hardware requirements for a deployment of IBM Cloud Pak for Watson AIOps on an OpenShift cluster. It includes the hardware requirements for IBM Cloud Pak for Watson AIOps, and for the OpenShift control plane and compute nodes.
This table is useful for sizing your environment when you need to create a new OpenShift cluster to deploy the full OpenShift and IBM Cloud Pak for Watson AIOps stack.
Note: vCPU is defined as when one x86 CPU splits each of its physical cores into virtual cores (vCPU). It is assumed that one x86 CPUs physical core can be split into two logical vCPUs.
| Category | Resource | Small | Large |
|---|---|---|---|
| Overall environment Resource count |
Node count vCPU Memory (GB) CoreOS root disk(GB) Persistent storage (Gi) |
6 84 240 720 1,549 |
11 188 432 1320 2,642 |
Hardware requirements for IBM Cloud Pak for Watson AIOps services only
The following table shows the hardware requirements for a deployment of IBM Cloud Pak for Watson AIOps on an OpenShift cluster. It includes the hardware requirements for IBM Cloud Pak for Watson AIOps, but not for the OpenShift control plane and compute nodes.
This table is useful for sizing when an OpenShift cluster is already installed, but extra nodes are required to accommodate IBM Cloud Pak for Watson AIOps.
| Category | Resource | Small | Large |
|---|---|---|---|
| Overall environment Resource count |
Node count vCPU Memory (GB) CoreOS root disk(GB) Persistent storage (Gi) |
3 72 192 360 1,549 |
8 176 384 960 2,642 |
Resource allocation for the OpenShift nodes
These tables give the recommended resource allocation for the OpenShift master and worker nodes, along with the recommended configuration for the disk volumes associated with each persisted storage resource.
Hardware requirements
| Category | Resource | Small | Large |
|---|---|---|---|
| OpenShift control plane master resources per node | Node count vCPU Memory (GB) Disk (GB) |
3 4 16 120 |
3 4 16 120 |
IBM Cloud Pak for Watson AIOps components - suggested configuration
| Category | Resource | Small | Large |
|---|---|---|---|
| OpenShift worker resources per node | Node count vCPU Memory (GB) Disk (GB) |
3 24 64 120 |
8 22 48 120 |
| IBM Cloud Pak for Watson AIOps persistent storage minimum requirements (Gi) |
MongoDB Flink job manager recovery Cassandra CouchDB Elasticsearch Kafka Evtmanager Topology File Observer Zookeeper Elasticsearch ES Aiops Strimzi Kafka Strimzi Zookeeper Zen Metastoredb IBM Elasticsearch data Evtmanager-DB2ese Evtmanager Impact UI Evtmanager NCOBackup Evtmanager NCOPrimary Evtmanager-Openldap Minio Evtmanager NCIServer Postgres Keeper User Home Image registry storage Flink Zookeeper |
60 3 100 20 75 150 5 5 150 50 50 30 100 100 5 5 5 1 400 5 20 10 200 0 |
60 3 300 60 225 300 5 15 250 150 150 30 300 100 5 5 5 1 400 5 60 10 200 3 |
Hardware requirements for Metric Manager
If you are also installing on-premises Metric Manager, then you require the hardware in the following table, which must have Red Hat Enterprise Linux 7.x.
The number of key performance indicators (KPIs) that can be processed for each deployment size is shown, for an aggregation period of 5 minutes and a training period of 4 weeks.
| Deployment size | KPIs | Nodes | Resource | CPU (Cores per node) | Memory (GB) | Disk (GB) |
|---|---|---|---|---|---|---|
| Trial/Proof of concept (POC) | 60,000 | 1 | Server 1 (Analytics, Database, UI) | 8 | 36 | 200 |
| Small | 100,000 | 2 | Server 1 (Analytics) Server 2 (Database, UI) |
10 4 |
43 20 |
400 250 |
| Medium | 250,000 | 3 | Server 1 (Analytics) Server 2 (Database) Server 3 (UI) |
20 6 4 |
77 38 8 |
700 300 250 |
| Large | 500,000 | 3 | Server 1 (Analytics) Server 2 (Database) Server 3 (UI) |
24 8 4 |
147 54 8 |
1200 450 250 |