KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Data Lake
Centralized storage repository designed to hold large amounts of raw data in their native format. It allows storage of structured, semi-structured, and unstructured data at petabyte scale.
Data Swamp
Data Lake that has lost its governance and organization, making data difficult to access and use. It results from a lack of metadata management and proper documentation.
Data Lakehouse
Hybrid architecture combining the advantages of Data Lakes and data warehouses to provide unified data management. It enables direct analysis on data stored in an open format optimized for performance.
Data Ingestion
Process of collecting and transferring data from various sources to a centralized storage system like a Data Lake. It can be performed in real-time, batch, or streaming according to business needs.
Schema-on-Read
Approach where data structure is applied at the time of reading rather than writing. It offers maximum flexibility for storing heterogeneous data without defining a schema in advance.
Schema-on-Write
Traditional methodology where data schema must be defined before writing into the system. It ensures data quality and consistency but reduces storage flexibility.
Data Catalog
Organized and indexed metadata describing available data in a Data Lake. It facilitates data discovery, understanding, and governance through a centralized interface.
Data Governance
Set of policies, procedures, and standards defining data management within the organization. It ensures quality, security, compliance, and appropriate use of Data Lake data.
Data Partitioning
Technique for dividing data into smaller segments based on specific criteria such as date or category. It optimizes query performance by limiting reads to relevant partitions.
Data Sharding
Horizontal partitioning of data distributed across multiple servers to improve scalability and performance. Each shard contains a unique subset of the total data.
Data Replication
Process of copying data from one location to another to ensure high availability and fault tolerance. It can be synchronous or asynchronous depending on consistency requirements.
Data Versioning
Mechanism for tracking and managing data changes over time in a Data Lake. It facilitates auditing, error recovery, and temporal trend analysis.
Data Lineage
Complete traceability of the data lifecycle from source to final destination. It documents transformations, movements, and relationships between different data entities.
Data Mesh
Decentralized data management architecture treating data as distributed products. It eliminates bottlenecks from central teams by promoting functional domain autonomy.
Delta Lake
Open-source storage layer bringing ACID transactions to Data Lakes built on distributed file systems. It enables updates, deletions, and time travel queries on parquet data.
Apache Iceberg
Open-source table format for large analytical Data Lakes, offering optimal query performance and seamless schema evolution. It separates planning operations from execution operations.
Apache Hudi
Streaming data framework providing batch and real-time processing capabilities on Data Lakes. It enables incremental updates and deletions with consistency guarantees.
Data Virtualization
Data integration approach allowing access and manipulation of data without physically moving it from their sources. It creates a unified and abstract view of distributed data.
Data Fabric
Unified and intelligent data management architecture facilitating data access wherever it resides. It seamlessly combines data integration, governance, and orchestration.
Zone Medallion
Data Lake architecture organizing data into three zones: Bronze (raw), Silver (cleaned), and Gold (aggregated). It progressively structures data for analysis and decision-making.