Honeycomb for observability

Pick 3 of 4 for observabilityOfficialHoneycomb

Honeycomb's official server is our third pick for observability, and its strength is specific: slicing high-cardinality events to find the one outlier a dashboard would hide. When the question is why a particular subset of requests is slow, Honeycomb's query model is built for exactly that kind of unplanned investigation.

It ranks third of four because it is trace- and event-centric rather than a full-platform vendor, so metrics-and-dashboard breadth is not where it leads. Datadog, Grafana, and Prometheus each cover a different observability shape, and for some teams one of those is the more complete home.

How Honeycomb fits

The work runs on Honeycomb's query and trace tools. run_query executes time-series aggregations with calculations and breakdowns, and run_bubbleup identifies the dimensions that statistically separate the slow or failing events from the rest, which is how you find the outlier without guessing. get_trace renders a full trace as a waterfall, list_spans ranks span names by count, and get_span_details returns common attribute values per span. find_columns and get_dataset_columns locate the fields to break down on, while get_workspace_context and get_environment orient the agent in the right dataset.

Honest limits: this is not a single pane over metrics, logs, and dashboards, and it leans on instrumented events with useful cardinality to shine. Datadog is the stronger pick when you want one full-platform vendor across APM, logs, and metrics; Grafana fits teams on the open dashboard-and-Prometheus stack; Prometheus is the lean PromQL-only choice. Reach for Honeycomb when arbitrary-dimension event analysis is the job and run_bubbleup is the tool you keep wanting.

Tools you would use

ToolWhat it does
get_workspace_contextProvides team name, current time, and the environment list with dataset counts.
get_environmentReturns details for a single environment, including its datasets and calculated fields.
get_datasetRetrieves metadata and the full column schema for a dataset.
get_dataset_columnsReturns the column schema, with optional sample values for specific columns.
run_queryExecutes time-series aggregation queries with support for calculations and breakdowns.
get_query_resultsFetches results from a previously executed query by URL or ID.
find_queriesSearches query history and saved queries by intent.
find_columnsLocates columns and calculated fields using natural-language keywords.
run_bubbleupRuns BubbleUp analysis to identify statistically significant differences in query results.
get_traceRetrieves all spans for a trace ID and renders them as a waterfall.
Full Honeycomb setup and config →

FAQ

What does Honeycomb do that a dashboard tool does not?
It slices high-cardinality events to isolate an outlier. run_query breaks results down by arbitrary dimensions and run_bubbleup flags which dimensions statistically separate the bad events from the good ones, surfacing causes a fixed dashboard would miss.
Is Honeycomb a full replacement for Grafana or Datadog?
Not quite. It is event- and trace-centric rather than a single pane over metrics, logs, and dashboards. Grafana fits the open Prometheus stack and Datadog covers full-platform APM; Honeycomb leads on arbitrary-dimension event analysis.