Highly Available ArcGIS Enterprise deployment

Provision a highly available ArcGIS Enterprise deployment on three (or more for spatiotemporal data store) Amazon Web Service instances.

This template provisions a highly available base ArcGIS Enterprise deployment on two AWS instances plus one instance for a file server. A base ArcGIS Enterprise deployment includes Portal for ArcGIS, a GIS Server (used as the portal's hosting server), ArcGIS Data Store (which will create a relational and tile cache data store), and ArcGIS Web Adaptor. You can optionally direct the template to create a spatiotemporal big data store on additional instances. The template will set up an architecture as shown in the graphic below.

Highly available base ArcGIS Enterprise deployment

To deploy this template, follow these steps:

  • Create an S3 bucket in your AWS account to store Esri authorization files and your SSL certificate file.
    • Create an S3 bucket in your AWS account. This bucket is called DeploymentBucket in the template parameters.
    • Upload your Portal for ArcGIS license file and ArcGIS GIS Server license file to the DeploymentBucket.
    • Obtain the domain for your ArcGIS Enterprise deployment and an SSL certificate file issued to your domain name.
    • Upload your SSL certificate into AWS Certificate Manager. Write down the Amazon Resource Name (ARN) of the certificate.

  • Run the template.
    Click Launch Stack for this template. There are several parameters that can be set when launching the stack using the template, such as the following:
    • ASInstanceType: The EC2 instance type for machine 1 and machine 2, as shown in the graphic. They are the EC2 instances that participate in the ArcGIS Enterprise deployment. They will be configured with the AWS Auto Recovery feature. If the instance crashes, AWS can restore it in the same Availability Zone to the stage before it crashed. Only certain instance types are allowed.
    • BDSInstances: AWS instance for the spatiotemporal data store. Pick 0 or 1 to create AWS instance for spatiotemporal data store or not. The CloudFormation stack will create an Auto Scaling group. This allows you to modify the number of instances in this Auto Scaling group in AWS Management Console after the stack is created. For example:
      • Pick 0 if you do not want a spatiotemporal big data store when creating a base ArcGIS Enterprise stack.
      • If you later want a spatiotemporal big data store, go to the AWS Management Console, find the Auto Scaling group that has your CloudFormation stack name, and change the Desired, Min, Max, parameter values to 1.
      • Pick 1 if you want a spatiotemporal big data store.
      • If you want more than 1 instance in the spatiotemporal big data store, always make sure to change the Desired, Min, and Max number to 1 first. Wait until the spatiotemporal big data store is registered to your base ArcGIS Enterprise deployment's hosting server successfully. Then change the Desired, Min, and Max number to the number you want. For example, if you want the spatiotemporal big data store to contain three instances, set Desired, Min, and Max to 3.
    • BDSInstanceType: The instance type you want your spatiotemporal big data store to be.
    • DeploymentBucket: The S3 bucket you created. It has your Esri authorization files and SSL certificate file uploaded to it.
    • DriveSizeRoot: The size of the root drive. It's the C: drive on Windows and the root drive on Ubuntu Linux.
    • ELBName: Name of an existing ELB or specify 'NEW_ELB' to create and use a new one.
    • FSInstanceType: The instance type for the file server machine. This is machine 3 in the graphic. This machine hosts the server directories for the GIS Server site (and the ArcGIS Server configuration store and Portal for ArcGIS content store if you picked "CloudStore" for "StoreType" parameter). This machine will be configured with the AWS Auto Recovery feature. If the instance crashes, AWS can restore it in the same Availability Zone to the stage before it crashed. Only certain instance types are allowed.
    • KeyName: The KeyPair to allow you remote access to the instances.
    • PortalLicenseFile: The authorization file for Portal for ArcGIS that you obtained from Esri. It must be in the S3 deployment bucket.
    • PostInstallationScript: ZIP archive file with custom post installation script. This zip file must be uploaded to DeploymentBucket. If there's a deploy.bat or deploy.sh at the root level inside this zip file, deploy.bat will be executed as administrator on Windows as the last step to configure the machine. On Ubuntu, deploy.sh will be executed as superuser.
    • RunAsUserPassword : This is the password for the account used to run the ArcGIS services. (Windows only)
    • ServerLicenseFile: The authorization file for ArcGIS GIS Server that you obtained from Esri. It must be in the S3 deployment bucket.
    • SiteAdmin: The user name for the Portal for ArcGIS initial administrator account and ArcGIS Server primary site administrator.
    • SiteAdminPassword: The password for SiteAdmin.
    • SiteDomain: The domain name for your ArcGIS Enterprise deployment. This domain must be mapped to the Elastic Load Balancer you specified.
    • StoreType:
      • FileSystem: The Portal for ArcGIS content store and ArcGIS Server configuration store will be on the file system in the same machine.
      • CloudStore: The Portal for ArcGIS content store will be in an AWS S3 bucket created by the stack. The ArcGIS Server configuration store will be in AWS DynamoDB and an S3 bucket created by this stack.
    • Subnet1: The id of the subnet you want this deployment to be deployed to.
    • Subnet2: The id of another subnet you want the second instance in this deployment to be deployed to.
    • VPCId: The id of the VPC where you want to deploy to. The Subnets must belong to this VPC.

  • Map your domain name to the ELB created with this stack.

    After the stack is created successfully, the output of this stack has the URLs to ArcGIS Server Manager and REST page using your domain name. There's also an output for the Elastic Load Balancer (ELB) created in front of this deployment. You need to map your domain name to the ELB. For example, map WebGIS.esri.com to  arcgis-WebGISha-123456789.us-east-1.elb.amazonaws.com through CNAME mapping.

    • Notes:

    In this template if you picked "CloudStore" for the "StoreType" parameter, the DynomoDB's provision capacity is set to:
    • Read Capacity units to 250 tables
    • Write Capacity units to 25 tables
    The estimated cost is about $36/month. Esri testing indicates this is good for publishing approximately 500 services. You can edit this setting in the AWS Management console to decrease the units to lower the cost, or increase the units for more services. Some ArcGIS Server functionalities, such as publishing, could fail if the provisioned capacity is not sufficient.

Troubleshooting

See Troubleshoot AWS CloudFormation stack creation in the ArcGIS Enterprise on Amazon Web Services help if you run into problems.