Understanding Edge Integration Cell Architecture

Objective

After completing this lesson, you will be able to describe the Edge Integration Cell architecture and explain how centralized management and decentralized execution separate data processing.

Introduction

Lesson 2.2 positioned Edge Integration Cell within SAP Integration Suite: what it is, what it supports, and when to choose it. This lesson goes deeper - into the architectural components that make EIC work.

Architecture is not just a diagram. It is the set of design decisions that determines what a system can do, how it behaves under different conditions, and where responsibilities lie. Understanding the EIC Architecture allows you to answer questions that matter in practice: Where does my data actually go? What happens if the cloud connection is interrupted? Who manages which components? How can EIC run in multiple locations?

The central architectural principle of Edge Integration Cell is centralized management and decentralized execution.

The Two-Plane Model

The Two-Plane Model

Edge Integration Cell is built on a clear architectural split between two planes:

PlaneLocationManaged ByPurpose
Control PlaneSAP Business AI Platform (Cloud)SAPDesign, deployment orchestration, monitoring, security material management
Data PlaneCustomer's Kubernetes cluster (Private)Customer (infrastructure); SAP (EIC software layer)Message processing - where integration flows actually run

This separation is the architectural cornerstone of EIC. It allows organizations to leverage the governance capabilities of a cloud platform while keeping actual data processing entirely within their own security perimeter.

Analogy: Think of a large international bank. The bank's headquarters sets the rules, defines procedures, monitors performance, and manages compliance across all branches. Each regional branch serves customers locally, transactions happen on-site, with the money and client data never leaving the branch building. Headquarters has full visibility into what is happening at every branch, but the actual transactions stay local. This is precisely the EIC model: centralized control, local execution. (Note: As with any analogy, this is a simplification. When Trace-level logging is enabled for troubleshooting, message payload data can be transmitted to cloud monitoring, see the Data Separation section below for details.)

Centralized cloud management with decentralized local execution

Control Plane: Centralized Management

The control plane is the cloud-based side of the EIC architecture. It runs within SAP Business AI Platform and is accessed through the SAP Integration Suite Cockpit. Regardless of whether integration flows will eventually execute in the cloud or locally on an EIC instance, all design and management work happens here.

Key functions of the control plane:

  • Integration Flow Design: The SAP Integration Suite Cockpit provides a browser-based environment for creating, configuring, and versioning Integration Flows and API proxies.
  • Integration Flow Design: The SAP Integration Suite Cockpit provides a browser-based environment for creating, configuring, and versioning Integration Flows and API proxies.
  • Deployment Orchestration: When an integration artifact is ready, deployment instructions are sent from the control plane to the target runtime. The control plane tracks deployed versions and enables controlled updates.
  • Monitoring and Observability: Dashboards in the SAP Integration Suite Cockpit display Message Processing Logs, error reports, and performance metrics - for all runtimes, including remote EIC instances. Operations teams do not need separate monitoring tools for cloud and EIC scenarios.
  • Security Material Management: Credentials, certificates, and security artifacts are managed centrally in the cloud and distributed securely to the runtimes that need them.

Critical constraint: Under normal operating conditions (log level Info or Error), no business message content passes through the cloud control plane. Only metadata, configuration data, and monitoring information are exchanged between SAP Business AI Platform and the EIC runtime. The actual content of integration messages - production orders, quality inspection records, customer data - stays within the private network. Note: When the Trace log level is enabled for a specific integration flow, payload data may be included in log records accessible from cloud monitoring. This is a documented exception relevant for compliance planning.

Data Plane: Decentralized Processing

The data plane is the customer-managed side of the EIC architecture - the local environment where integration scenarios actually execute. It consists of the EIC runtime components deployed on a Kubernetes cluster within the customer's private infrastructure.

Key characteristics of the data plane:

Local Message Processing: Integration flows and API proxies execute within the private network. Sensitive business data remains within the customer's defined security boundaries.

No Internet Traversal for Business Data: Messages between connected systems flow through the EIC runtime without leaving the corporate network.

Kubernetes Foundation: The EIC runtime is packaged as containerized workloads deployed on Kubernetes. Kubernetes provides container orchestration: scheduling workloads, managing container lifecycles, handling restarts. Unit 3 covers Kubernetes in depth.

Customer-Managed Infrastructure: The customer provisions, maintains, and operates the underlying Kubernetes cluster. SAP manages the EIC application layer.

Architecture Components Reference

The following table describes the key components in the EIC architecture:

ComponentLocationRole
SAP Integration SuiteSAP Business AI Platform (Cloud)Design environment, deployment control, monitoring dashboards, security material management. The single management interface for all runtimes.
Kubernetes ClusterCustomer private infrastructureThe container orchestration platform on which EIC runs. Provisioned and operated by the customer. Supported platforms listed in Lesson 2.6.
Edge Lifecycle Management (ELM)Customer private (ELM Bridge on cluster) + SAP Business AI PlatformInstalls, updates, and monitors the EIC runtime. Manages the connection between the Kubernetes cluster and SAP Integration Suite. Primary tool for EIC infrastructure administration.
Cloud ConnectorCustomer private (embedded in ELM by default)Establishes the secure outbound tunnel from the private infrastructure to SAP Business AI Platform. By default, an embedded Cloud Connector is deployed and managed through Edge Lifecycle Management as part of the EIC setup process. Customers who prefer to use an existing, external Cloud Connector installation can configure EIC to use that instance instead.
PostgreSQL or SAP HANACustomer privateRelational database used by the EIC runtime for Integration Flow state persistence, message store, and Message Processing Logs (MPLs).
Redis or ValkeyCustomer privateIn-memory data store used for messaging and caching within the EIC runtime.

Note

Cloud Connector: The communication between the EIC runtime and SAP Business AI Platform is established through the Cloud Connector. The connection is always initiated from the private side (outbound from the customer's network). No inbound firewall ports to the customer's infrastructure are required for EIC's operational communication. Lesson 2.4 covers the communication model in detail.

Data Separation

What Flows Where?

The two-plane model raises a natural question: if management happens in the cloud and execution happens locally, what data actually crosses the boundary? Understanding this separation is essential for compliance and security conversations. EIC distinguishes three categories of data, each with a clear rule for where it flows:

  • Business messages - stays in private infrastructure only:
    • The body/payload of Integration Flow messages - production orders, quality records, financial data, customer data, personal data
    • Processed entirely within the data plane; never transmitted to or persisted in SAP Business AI Platform (Exception: When MPL Attachments or Trace Log Level is configured, message content may be included in MPLs and become readable in the SAP Integration Suite monitoring UI → see Key principle below)
  • Management data - synchronized between cloud and EIC:
    • Integration artifact definitions (iFlow metadata and configuration)
    • Security material references (credential references, not credential values)
    • Lifecycle management commands (deploy, undeploy, update)
    • Software update packages
  • Message Processing Logs (MPLs) - generated locally, queryable on-demand via cloud when connected:
    • MPLs are generated asynchronously by the EIC runtime: both the Worker and the Policy Engine write log entries to the Message Service, from where the monitoring data consumer persists them to the local database (PostgreSQL or SAP HANA)
    • They contain operational metadata: processing timestamp, status (success/failed), error classification, processing duration
    • By default, MPLs do not contain business message content. When MPL Attachments or Trace Log Level is configured, message content may be included
    • MPLs are not transmitted to the cloud. The SAP Integration Suite Cockpit queries them on-demand via an API request through the Cloud Connector and displays them in the Message Monitor. No MPL data is persisted in the cloud

    Key principle: This separation allows EIC to simultaneously satisfy data sovereignty requirements and enable centralized monitoring. Under the default log level (Info/Error), business message content remains exclusively within the private network. When the Trace log level is enabled, payload data is included in MPLs and becomes readable in the SAP Integration Suite monitoring UI via on-demand query.

Multi-Instance Deployments

Multiple Runtime Locations: Up to 20 EIC Instances

Edge Integration Cell supports deployment across multiple runtime locations. An SAP Integration Suite tenant can register and manage up to 20 Edge Integration Cell instances. This enables scenarios such as:

  • One EIC instance per geographic region (Frankfurt, Singapore, São Paulo), each processing integration traffic locally
  • One EIC instance per manufacturing plant or production facility
  • One EIC instance for production, one for a quality assurance environment

At deployment time, integration developers select the specific runtime location (EIC instance) where an artifact should execute. Different artifacts within the same tenant can run on different EIC instances - or on the Cloud Integration Runtime - according to their requirements.

Centralized Management of Multiple EIC Runtime Locations.

Scenario Check-In: Global Manufacturing AG

For Global Manufacturing AG, the architecture maps directly to their Frankfurt data center. Their SAP Integration Suite tenancy manages the EIC instance registered as "FRA-EIC-PROD," running on a Kubernetes cluster in Frankfurt. When the S/4HANA-to-QMS iFlow executes, quality inspection messages flow from S/4HANA through the local EIC runtime to the QMS - entirely within Frankfurt. Monitoring data flows back to the cloud Cockpit, giving the integration team full operational visibility without any business data leaving Germany.

Key Takeaways

Edge Integration Cell follows the principle of centralized control and decentralized execution. SAP Integration Suite (cloud) manages design, deployment, and monitoring. The Kubernetes cluster - together with a local database, in-memory store, Cloud Connector, and Edge Lifecycle Management - executes integration logic locally. Under default logging settings, business messages never leave the private infrastructure. Up to 20 EIC instances can be managed from a single SAP Integration Suite tenant.

Further Resources

SAP Help Portal – Operating Model for Edge Integration Cell

https://help.sap.com/docs/integration-suite/sap-integration-suite/operating-model

SAP Help Portal – Technical Landscape for Edge Integration Cell

https://help.sap.com/docs/integration-suite/sap-integration-suite/technical-landscape-edge-integration-cell

SAP Architecture Center – Edge Integration Cell on Hyperscalers

https://architecture.learning.sap.com/docs/ref-arch/0a229d