Open Edge Platform System Requirements#

Edge Orchestrator Cloud-based AWS Resource Quotas#

Many AWS resources are restricted by per-account service quotas.

The following lists the quantity of service quota limited resources required for each Edge Orchestrator instance installed. Ensure that the service quota settings accomodate the allocation of the following resources. To request a service quota increase, see Requesting a quota increase.

  • Elastic IP addresses (EIPs) - 7 EIPs required per Edge Orchestrator instance. The quota is specified per region and defaults to five. This has to be increased to support an Edge Orchestrator installation.

    • 3 For NAT Gateway that allows outbound internet access for the Edge Orchestrator instance.

    • 3 For the Network Load Balancer that allows inbound access to the Edge Orchestrator BIOS service.

    • 1 For the bastion host that allows to access private API endpoints such as AWS EKS* and AWS RDS*.

  • Application Load Balancers (ALBs) - Two ALBs are required per Edge Orchestrator instance. This quota is applied per region.

  • Network Load Balancers (NLBs) - One NLB is required per Edge Orchestrator instance. This quota is applied per region.

  • Amazon Virtual Private Cloud* (Amazon VPC*) - One Amazon VPC is required per Edge Orchestrator instance. This quota is applied per region.

  • NAT gateways in each availability zone (AZ) - One NAT gateway per AZ (three AZs specified for each deployment) is required for each Edge Orchestrator instance. This quota is applied per AZ.

  • Amazon Relational Database Service* (Amazon RDS*) DB instances - Three Amazon RDS DB instances are required. This quota is applied per region.

  • Amazon Simple Storage Service* (Amazon S3*) buckets - 11 Amazon S3 buckets required per Edge Orchestrator instance. The quota is specified per account and must accomodate resources required in all regions.

  • Amazon Elastic File System (Amazon EFS) - There is a single quota applied per account that needs to specify the total available Amazon EFS space. The value set in this quota must allow for each Edge Orchestrator instance to use up to approximately 50 GB.

AWS Resource Requirements#

The resource requirement depends on the number of edge nodes connected to the Edge Orchestrator. Intel has validated the following configurations for various scales. The following Resource requirements can be changed during installation script. The –profile options allows you to decide which scale profile they want to use. Currently, there are four profiles:

  • default(3 x t3.2xlarge + 1 x t3.2xlarge nodes)

  • 100 edge node(3 x t3.2xlarge + 1 x r5.2xlarge)

  • 1k edge node (3 x m4.4xlarge + 3 x r5.4xlarge)

More options allows you to fine tune your cluster scale abilities:

  • –min-o11y-nodes, –desired-o11y-nodes, –max-o11y-nodes: sets min, desired, and max number of observability nodes. Refer to Observability Configuration for guidance.

  • –min-rds-acu, –max-rds-acu: sets min and max Aurora Capacity Unit(ACU)

  • –node-type: sets the default worker node type (for example, m4.4xlarge)

  • –o11y-node-type: sets the observability worker node type (for example, r5.2xlarge). Refer to Observability Configuration for guidance.

Refer to Start Edge Orchestrator Installation Environment to use these options during the installation process and Manual Installation - Terraform Configuration to learn more about how to configure scale-related parameters in the configuration files.

Minimum Requirement (up to 100 Edge Nodes)#

  • Multi-node Amazon EKS cluster

    • Three-node Amazon Elastic Kubernetes Service (Amazon EKS*) node group backed by three t3.2xlarge instances (refer to Amazon EC2 Instance types)

      • CPU: Eight vCPUs (each) on 2.4 GHz Intel® Xeon® E5-2676 v3 Processor

      • RAM: 32 GiB (each)

      • Disk: 128-Gb elastic block storage (EBS)

    • One-node Amazon EKS node group backed by a r5.2xlarge instance (Refer to Amazon EC2 Instance types)

      • CPU: 8 vCPUs on 3.1-GHz Intel® Xeon® Platinum 8175 Processor

      • RAM: 64 GiB

      • Disk: 128-Gb EBS

      • This node is dedicated to the Observability stack. See Observability Configuration for details.

  • Database: Two Amazon* Aurora capacity units (ACUs) of SQL database.

Common Requirement Across Various Scales#

  • Storage: Auto-scaled through Amazon EFS, from 16 GiB. Observability stack uses Amazon EBS.

  • Network: Default Amazon VPC

Edge Orchestrator Observability Configuration#

Observability components are deployed on a dedicated observability node group, which by default contains a single t3.2xlarge instance.

For larger deployments, you must scale up the capacity according to the recommendation.

Number of edge nodes that can be handled by specific instances in observability node group are as follows:

Edge Nodes (up to)

Observability Profile

Pod-configs profile

Observability Node Type

Observability Node Count

50

o11y-release

default

t3.2xlarge

1

100

o11y-release

100en

r5.2xlarge

1

500

o11y-release-large

500en

r5.4xlarge

2

1000

o11y-release-large

1ken

r5.4xlarge

3

Note

In case of deployment beyond 1k edge nodes it is recommended to use multiple r5.8xlarge instances (preferred observability node count is the multiple of 3).

Due to the limit of EBS volumes attached to a single EC2 instance, it is recommended to use two or more observability nodes for the o11y-release-large observability profile (Refer to Amazon EBS volume limits for Amazon EC2 instances). Note that placing multiple observability nodes in the same availability zone (AZ) will reduce AWS regional data transfer costs.

See also Scale Edge Orchestrator for more information about pod-configs profiles and how to scale the cluster.

The default replication factor for edge node logs and metrics is 3. You can configure it in the cluster definition with these settings: .Values.argo.o11y.edgeNode.loki.replicationFactor and .Values.argo.o11y.edgeNode.mimir.replicationFactor.

Note

Replication factor 2 is not supported for Grafana Mimir* storage (metrics).

The default data retention period for edge node logs and metrics is 15 days (360h). You can configure it in the cluster definition with these settings: .Values.argo.o11y.edgeNode.loki.logRetentionPeriod and .Values.argo.o11y.edgeNode.mimir.structuredConfig.metricsRetentionPeriod.

The default data retention period for edge node provisioning logs is 31 days (744h). You can configure it in the cluster definition with this setting .Values.argo.o11y.edgeNode.loki.provisioningLogRetentionPeriod.

Note

The minimum data retention period for edge node logs and metrics is 1 day (24h).

Supported Operating Systems and Browsers#

You can access Edge Orchestrator through the web browser; it does not require local installation, therefore there is no specific OS version.

The supported browsers are as follows:

  • Chrome*

  • Edge*

  • Safari*

  • Firefox*

For the best Edge Orchestrator experience, use the Chrome browser.