Analysis of Integrations in SAP Cloud Integration: Exploring Patterns and Detecting Anomalies

Event: SAP Inside Track Madrid 2026

Date: April 17, 2026

Speaker: Vadim Klimov

Session: Analysis of integrations in SAP Cloud Integration - Exploring patterns and detecting anomalies

Transcript

1. Analysis of Integrations in SAP Cloud Integration Exploring Patterns and Detecting Anomalies Vadim Klimov SAP Integration Architect SAP Inside Track Madrid April 17, 2026
2. Speaker Info Integration architect and developer | SAP Cloud technologist | SAP BTP · AWS · Azure Author | SAP PRESS / Rheinwerk Publishing Speaker | SAP technology events /vadim-klimov Dr. Vadim Klimov
3. The Problem Question 1: Do any iFlows use basic authentication for HTTPS calls to external systems? Question 2: Are there any iFlows that use unusual PGP encryption settings? Question 4: Do any iFlows feature anomalous parallelization settings in their splitter steps? Question 5: How many iFlows use an outdated version of the Advanced Event Mesh sender adapter? Question 6: Which iFlows targeting a service at https://demo.dev use rare retry configuration in a receiver adapter? Question 3: Which iFlows poll files from the SFTP server demo.dev and will be affected by its decommissioning?
4. How much time do you think it will take to answer these questions?
5. The Problem | Time To Answer iFlow complexity iFlow count 20-25 low to medium (mostly linear flow logic, less than 15 steps per iFlow)
6. The Problem | Time To Answer iFlow complexity iFlow count 1000 medium to high (modularization, branching logic, multiple external calls, more than 25 steps per iFlow)
7. Intelligent Search and Analysis Tools Machine Learning Unsupervised Transductive Anomaly Detection Generative AI Retrieval-Augmented Generation (RAG) SAP Cloud Integration | Analysis of Integrations Governance and Quality Assurance Conversational Search Anomaly Detection Validate integrations against best practices and design guidelines managed via Docs-as-Code. Query and analyze integration definitions and configurations using natural language prompts. Identify anomalies and atypical patterns within integration definitions and configurations.
8. Anomaly Detection
9. SAP Cloud Integration Integration flow Processing layer Anomaly detection engine Key Components | Overview Metadata Storage layer Filesystem, Git-managed repository Load artifact data Definition Participants, connections, flow steps Resources Scripts, schemas, mappings, archives Configuration Externalized parameters configuration Designtime artifacts Stores Reporting layer Analysis summary and detailed report API Security artifacts Fetch and load artifact data References Fetch artifacts
10. Anomaly Detection | Pipeline Data ingestion SAP Cloud Integration API consumption (real-time integration with tenant’s workspace) or filesystem access (tenant’s workspace snapshot) to retrieve information about integration flows (metadata, definitions and configurations). Data preprocessing Feature engineering Inference and anomaly detection Post-processing and evaluation Deserialization and parsing of integration flow definitions, resolution of configurations (externalized parameters), noise reduction (high-cardinality filtering), structural harmonization and normalization of flow elements’ properties, missing value imputation. Frequency encoding for categorical flow elements’ properties. Anomaly detection algorithm inference - currently, Copula-Based Outlier Detection (COPOD) is used. Some other algorithms - based on Isolation Forest (IF) and Histogram-Based Outlier Score (HBOS) - are considered. Two-tiered analysis: per-property profiling (detection of rare flow elements’ property values) and per-element scoring (detection of overall abnormality of each flow element). Thresholding, interpretation, feature contribution analysis, anomaly attribution.
11. What You See | iFlow Editor in SAP Integration Suite Web UI
12. What You Get | Anatomy of iFlow Artifact
13. What You Miss | Anomalous iFlow Definitions and Configurations