This topic explains the core terms and concepts of the UModel observability data modeling system. The terms are categorized by function to help you quickly understand and use UModel.
Core concepts
UModel
Definition: The Universal Observability Model. UModel is a data modeling method for observable data based on a graph model.
Description: UModel describes the IT world with a graph structure composed of nodes and links. This standard data modeling method provides a unified representation for observable data, decouples storage, and enables intelligent analysis.
EntitySet
Definition: An entity set. An entity set defines a collection of entities that share the same properties.
Description: An entity set is similar to a table schema in a database. It defines the structure, properties, and behavior of a class of entities, such as "apm.service" or "k8s.pod".
Entity
Definition: An entity instance is a concrete object in an EntitySet.
Description: A specific business object instance, such as the service instance "order-service" or the pod instance "web-pod-001".
EntitySetLink
Definition: An entity relationship definition. An entity relationship definition describes the type of relationship between two entity sets.
Description: An EntitySetLink defines the relationships between entities, such as "calls", "contains", or "runs_on". It is a core component for building an Entity Relationship Diagram.
EntityRelation
Definition: An entity relationship instance. An entity relationship instance is the specific relationship data between entities.
Description: This is the actual entity relationship data that describes the connection between specific entities, such as "user-service -> web-pod-123" (the service runs on the pod).
Data modeling
Field
Definition: A field. A field is the most basic unit of data in UModel.
Description: A field defines the specific properties of data, such as logs, metrics, traces, and entities. It contains complete metadata, including its name, type, constraints, and display characteristics.
TelemetryDataSet
Definition: An observable data set is the core concept for defining and managing observable data.
Description: A TelemetryDataSet provides a unified data modeling framework that supports multiple types of observable data, such as logs, metrics, traces, events, and profiles. It includes time dimensions and observation characteristics.
MetricSet
Definition: A metric set. A metric set defines the data structure for metrics.
Description: A MetricSet is used to describe a collection of metrics that share the same properties. It supports tag management, metric aggregation, and query optimization.
LogSet
Definition: A log set. A log set defines the structure of log data.
Description: A LogSet inherits the basic capabilities of TelemetryDataSet. It is suitable for modeling various types of log data and offers maximum flexibility.
TraceSet
Definition: A trace set. A trace set defines distributed tracing data.
Description: A TraceSet includes fields and protocol support specific to tracing. It is the abstract representation of tracing standards such as OpenTelemetry within UModel.
EventSet
Definition: A link set defines the event data structure.
Note: This applies to scenarios involving alert events, system events, and business events. The format follows the CloudEvents specification.
ProfileSet
Definition: A profile set. A profile set defines performance profiling data.
Description: A ProfileSet supports the modeling of various performance data, such as CPU, memory, and network data. It is mainly used for application performance analysis and system performance tuning.
Storage-related classes
Storage
Definition: Storage. Storage is an abstract concept in UModel that defines the physical storage of data.
Description: Storage decouples data modeling from specific storage implementations. It supports multiple storage classes and provides a unified storage interface.
EntityStore
Definition: Entity Storage is the unified storage engine in the UModel system.
Description: EntityStore stores and manages core metadata in the observability domain, including UModel metadata, entity data, and entity relationship data.
Simple Log Service Logstore
Definition: A Simple Log Service (SLS) Logstore. A Simple Log Service Logstore is a storage class for log data.
Description: It provides high-performance log writing and retrieval capabilities. It also supports full-text search and real-time analysis.
SLS MetricStore
Definition: An SLS MetricStore. An SLS MetricStore is designed for storing time series data.
Description: It provides efficient metric aggregation and query capabilities. It is optimized for storing and querying time series data.
Managed Service for Prometheus
Definition: Managed Service for Prometheus storage. This is a managed Prometheus service storage.
Description: It supports standard Prometheus Query Language (PromQL) queries and is fully compatible with the Prometheus protocol. It provides enterprise-grade high availability and disaster recovery.
Relationships
DataLink
Definition: A data link. A data link defines the relationship between an EntitySet or Link and a DataSet.
Description: A DataLink connects the entity model and the data model. It associates entities with observable data.
StorageLink
Definition: A storage mapping. A storage mapping defines the mapping relationship between an EntitySet or DataSet and Storage.
Description: A StorageLink connects a dataset to the underlying storage. It establishes a correspondence between logical fields and storage fields through field mapping.
Queries
USearch
Definition: An entity retrieval engine that serves as the core query control of UModel.
Description: USearch supports multiple query modes, including full-text search, exact match, and conditional filtering. It provides powerful entity query capabilities.
Query features:
Cross-type joint search.
Multi-keyword search scoring.
Scan capability.
Structured Process Language (SPL) integration.
Graph Query
Definition: A graph query. A graph query is a complex query operation based on entity relationships.
Description: It supports graph-match and graph-call functions to query relationships and paths between entities.
Query methods:
graph-match: Graph pattern matching.
graph-call: Graph function invocation.
Cypher: A standard graph query language.
UModel Query
Definition: A metadata query is a dedicated interface that queries knowledge graph metadata.
Description: You can use the .umodel query syntax to explore EntitySet definitions, EntitySetLink relationships, and the complete knowledge graph structure.
SPL
Definition: The SLS Processing Language is the language used for data processing and analysis in SLS.
Description: For more information, see SPL syntax.
Managing operation classes
Workspace
Definition: A workspace. A workspace is the basic container in UModel.
Description: All UModel operations must be performed within a specific workspace. Workspaces provide data isolation and permission management.
CommonSchema
Definition: A common pattern. A common pattern is a library of standardized UModel templates provided by the Alibaba Cloud observability platform.
Description: CommonSchema is a collection of predefined UModel components. It includes standardized entity sets, datasets, and relationships that are ready to use out-of-the-box.
UModel Explorer
Definition: The graphical management tool for UModel.
Description: You can use UModel Explorer to view and edit various UModel elements. It provides a visual interface for modeling and management.
Technical implementation
Lifecycle methods
Definition: Methods for managing the state of entities and relationships.
Description: These methods control lifecycle actions such as creating, updating, expiring, and deleting entities and relationships.
Method types: Create, Update, Expire, Delete, and Revise.
Timestamp
Definition: A time marker used to record the time of a data record.
Format: UNIX timestamp in seconds.
Related fields:
__first_observed_time__: The time when the entity was first observed.__last_observed_time__: The time when the entity was last observed.__keep_alive_seconds__: The time to live (TTL).
Primary key fields
Definition: Fields that uniquely identify an entity.
Description: Primary key fields are configured using primary_key_fields. UModel supports single-field and composite-field primary keys. The primary key remains stable throughout the entity lifecycle.
Entity ID
Definition: A unique entity identifier. An Entity ID is a 128-bit hexadecimal string.
Generation methods:
Standard: Apply the MD5 or double xxhash function to the primary key.
Fallback: The system automatically performs an xxhash conversion.
Data formats
Semantic string
Definition: A semantic_string. A semantic string is a string type that supports multiple languages.
Format: {"zh_cn": "Chinese content", "en_us": "English Content"}.
System fields
Domain
Definition: The business domain to which an entity belongs.
Examples: apm, k8s, acs, infrastructure.
Kind
Definition: The type identifier for a UModel element.
Types: entity_set, metric_set, log_set, trace_set, entity_set_link, data_link, storage_link.