liveBook · Manning

Dimensional label

As explained in Designing AI Systems

A metric label is a named attribute attached to a metric that enables grouping and querying by dimensions such as provider, model, or workflow.

A dimensional label is a named attribute attached to a metric, allowing operators to aggregate and query telemetry by provider, model, workflow, application, or team. section 7.4.2 section 7.8

Dimensional labels turn one metric into targeted operational views:

  • Filter model cost by workflow and model to produce a focused cost report. section 7.4.2
  • Compare throughput, latency, errors, token use, cache hits, and guardrail triggers across models or providers. section 7.4.2
  • Aggregate costs and other telemetry into application- and team-level operational reports. section 7.8
  • Apply the same query dimensions to automatically collected platform metrics and developer-published custom metrics. section 7.5.4

Targeted cost query

A single MODEL_COST_USD metric can answer: “What was the total cost for the patient-intake workflow using GPT-4o over the past 30 days?” Filtering by workflow and model supplies the dimensional view without scanning individual log events. section 7.4.2

Listing 7.4 Platform metric definitions and naming conventions
class PlatformMetrics:
    """Standard metrics emitted by platform services.
    Names follow: ai.platform.{service}.{metric_name}"""
 
    # Model Service
    MODEL_REQUEST_TOTAL = \
        "ai.platform.models.requests_total"
    MODEL_REQUEST_DURATION = \
        "ai.platform.models.request_duration_ms"
    MODEL_TOKENS_PROMPT = \
        "ai.platform.models.tokens.prompt"
    MODEL_TOKENS_COMPLETION = \
        "ai.platform.models.tokens.completion"
    MODEL_COST_USD = \
        "ai.platform.models.cost_usd"
    MODEL_CACHE_HITS = \
        "ai.platform.models.cache_hits_total"    #A
    MODEL_FALLBACKS = \
        "ai.platform.models.fallbacks_total"
 
    # Data Service
    DATA_SEARCH_DURATION = \
        "ai.platform.data.search_duration_ms"
    DATA_RELEVANCE_SCORE = \
        "ai.platform.data.relevance_score"    #B
 
    # Session Service
    SESSION_CONTEXT_TOKENS = \
        "ai.platform.sessions.context_tokens"
    SESSION_TRUNCATIONS = \
        "ai.platform.sessions.truncations_total"
 
    # Guardrails Service
    GUARDRAIL_VIOLATIONS = \
        "ai.platform.guardrails.violations_total"    #C
    GUARDRAIL_LATENCY = \
        "ai.platform.guardrails.evaluation_duration_ms"
 
    # Tool Service
    TOOL_EXECUTIONS = \
        "ai.platform.tools.executions_total"
    TOOL_CIRCUIT_BREAKS = \
        "ai.platform.tools.circuit_breaks_total"

Browse the standardized metric catalog to see how service-qualified names support discoverable, collision-resistant metric queries. section 7.4.2

How Designing AI Systems covers it

How deep it goes
Explains labels through metric filtering, dashboard aggregation, cost attribution, and custom observability rather than implementation details.
Examples
Uses model-cost queries, throughput dashboards, latency distributions, cache ratios, and guardrail-trigger reports.
What you can do after
Diagnose production behavior by drilling aggregate metrics into provider, model, workflow, application, or team dimensions.
How it connects
Connects metrics with providers, models, workflows, dashboards, distributed tracing, and cost attribution.

This explainer was generated by Manning's knowledge graph from the text of Designing AI Systems. The section links point to the parts of the book it draws on.