Holds one or more KQL databases and is where Real-Time Intelligence in Microsoft Fabric keeps time-stamped streaming events for analysis, indexed by the moment they arrive.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Eventhouse in context, with comparison tables and the common traps.
Terms in this definition
- KQL
Kusto Query Language, used read-only to query Azure Data Explorer, Log Analytics and Microsoft Sentinel; log alert rules are written in it too.
- WHERE
Limits a SELECT, UPDATE or DELETE to just the rows meeting a condition. Omit it, and the statement hits every row.
- Real-Time Intelligence
Microsoft Fabric's workload for data in motion. Using eventstreams, eventhouses with KQL databases, the Real-Time hub, Real-Time Dashboards and Activator, it takes in events, processes, stores and analyses them, and acts on them.
- Microsoft Fabric
Analytics platform from Microsoft delivered as SaaS on top of OneLake, offering capabilities including Direct Lake and shortcuts.
- Streaming
Sending model output piece by piece as server-sent events. Only the delivery changes; the answer's content, completeness and cost stay the same.
Related terms
- Capacity Overview Events
Real-Time hub events emitted by a Fabric capacity. Twice a minute a Summary event reports smoothed CU use and how much throttling is happening, and a State event fires on pausing, resuming or overload; Activator can alert on them, or an eventhouse can keep them as history.
- Capacity Planner
In preview, this eventhouse option stops the eventhouse from suspending when it's idle by keeping it always on. You can also set a floor of CUs for each hour-long slot across the week, while autoscale continues to operate.
- Database shortcut
Read-only by design, this eventhouse KQL database references another KQL or Azure Data Explorer database and keeps in sync with it, with no ingestion of its own. You can adjust its principals, permissions and caching policy, but nothing else.
- KQL database
Lives within a Microsoft Fabric eventhouse and keeps tables that you query in KQL. One eventhouse may contain several of them, and each can hold many tables.
- OneLake availability
When switched on for a KQL database or table (it starts disabled), an Eventhouse exposes that data in OneLake in Delta format. Other Fabric engines, Direct Lake included, can then read it without anyone copying it.
- Operations agent
A preview item in Real-Time Intelligence that keeps watch over eventhouse or ontology data, compares it with business rules and either proposes or performs actions. Failed pipeline runs can also be handed to it from the Monitor hub for a read-only investigation.
- Partitioning policy
A KQL policy set on an eventhouse table to reorganise data extents once ingestion has finished, either by hashing one string or guid column or by splitting time into evenly sized datetime ranges. Without it, ingestion time determines grouping, and Microsoft Learn recommends using it in just a handful of particular query scenarios.
- Query acceleration
A policy on an eventhouse that keeps delta tables reached through OneLake shortcuts indexed and cached for a chosen number of days, so queries run nearly as fast as on ingested data without moving it out of OneLake. Because these shortcuts are still external tables, you can't use update policies or materialized views on them.