Understand architecture profiles
When you create an ArcGIS Enterprise organization, you choose architecture profiles to set the initial level of availability and performance of ArcGIS Enterprise on Kubernetes. The profiles you choose determine the amount of CPU, memory, and disk space you will need for the initial deployment, and how those needs change as you scale your deployment. There are two types of architecture profiles:
An availability profile sets the initial number of pod replicas for each workload in ArcGIS Enterprise. Having more replicas provides greater redundancy and improves availability.
A performance profile sets the initial amount of CPU and memory each pod requests. Service pods with more CPU and memory generally have better throughput and response times.
Required system resources
The total amount of CPU and memory required to deploy ArcGIS Enterprise depends on both the number of pod replicas and the system resources used by each pod. The system requirements lists the minimum required resources for each combination of availability and performance profile.
It is important to note that publishing dedicated services, scaling up pod replicas, or adding capabilities such as raster analytics will increase the resources required.
Scenarios
Choosing the right architecture profiles involves tradeoffs. The best choice will balance your need for redundancy and performance against the costs of deploying the necessary infrastructure. Use the scenarios below to help you choose the best architecture profiles for your organization.
Scenario: Demonstration environment
An organization wants to demonstrate the capabilities of ArcGIS Enterprise on Kubernetes. This non-production environment will only be in service for a few hours, so there is no need to ensure high availability. The system should have excellent performance to showcase its capabilities.
Availability profile: Development and test environments
Performance profile: Enhanced performance
Scenario: Internal team deployment
A city planning department has a mission that is internally focused and does not have a strict service level agreement (SLA). Communicating planned or unplanned outages is straightforward and there is a high degree of flexibility for planned outages. The department needs the least costly production environment that is suitable for predictable and stable loads.
Availability Profile: Standard high availability
Performance Profile: Standard performance
Scenario: Organization-wide deployment
A large company uses ArcGIS Enterprise to serve the needs of many users throughout the entire organization. It is acceptable to have planned downtime for maintenance outside of normal operating hours but users expect advanced communication for any downtime. Short periods of slow service responses may be acceptable at peak times as long as they happen just a few times a month.
Architectural Profile: Standard high availability
Performance Profile: Standard performance
Scenario: Public-facing deployment
A state government uses ArcGIS Enterprise to provide information to various agencies. The agencies are formal stakeholders who have an expectation of continuous availability. The system must support continuous operating hours, but a planned maintenance window is acceptable if it is communicated through official channels with sufficient notice. Performance SLAs are not defined, but stakeholders have an expectation of good performance at all times and there is no tolerance for even occasional slow service responses.
Architectural Profile: Standard high availability
Performance Profile: Enhanced performance
Scenario: Operational system
A large company uses ArcGIS Enterprise to processes incoming data, conduct scheduled analysis, and generate reports from that analysis. It is important that this system remain continuously available, and there is an SLA requiring a specific level of uptime. Performance fluctuations, however, are tolerable because users are not actively waiting for responses.
Architectural Profile: Enhanced high availability
Performance Profile: Standard performance
Scenario: Viral-potential public-facing deployment
A national emergency response center uses ArcGIS Enterprise to provide vital public safety information during natural disasters. During these events, the system is used by millions of people who expect fast performance. There may be periods of the year, however, where no changes other than routine maintenance may be undertaken. There are formal SLAs for both uptime and service response time.
Architectural Profile: Enhanced high availability
Performance Profile: Enhanced performance