Druid has a distributed architecture that is designed to be cloud-friendly and easy to operate. You can configure and scale services independently for maximum flexibility over cluster operations. This design includes enhanced fault tolerance: an outage of one component does not immediately affect other components.
The following diagram shows the services that make up the Druid architecture, their typical arrangement across servers, and how queries and data flow through this architecture.
You can deploy Druid services according to your preferences. For ease of deployment, we recommend organizing them into three server types: [Master](#master-server), [Query](#query-server), and [Data](#data-server).
### Master server
A Master server manages data ingestion and availability. It is responsible for starting new ingestion jobs and coordinating availability of data on the [Data server](#data-server).
Master servers divide operations between Coordinator and Overlord services.
#### Coordinator service
[Coordinator](../design/coordinator.md) services watch over the Historical services on the Data servers. They are responsible for assigning segments to specific servers, and for ensuring segments are well-balanced across Historicals.
[Overlord](../design/overlord.md) services watch over the Middle Manager services on the Data servers and are the controllers of data ingestion into Druid. They are responsible for assigning ingestion tasks to Middle Managers and for coordinating segment publishing.
A Query server provides the endpoints that users and client applications interact with, routing queries to Data servers or other Query servers (and optionally proxied Master server requests).
Query servers divide operations between Broker and Router services.
[Broker](../design/broker.md) services receive queries from external clients and forward those queries to Data servers. When Brokers receive results from those subqueries, they merge those results and return them to the caller. Typically, you query Brokers rather than querying Historical or Middle Manager services on Data servers directly.
[**Router**](../design/router.md) services provide a unified API gateway in front of Brokers, Overlords, and Coordinators.
The Router service also runs the [web console](../operations/web-console.md), a UI for loading data, managing datasources and tasks, and viewing server status and segment information.
### Data server
A Data server executes ingestion jobs and stores queryable data.
[**Historical**](../design/historical.md) services handle storage and querying on historical data, including any streaming data that has been in the system long enough to be committed. Historical services download segments from deep storage and respond to queries about these segments. They don't accept writes.
[**Peon**](../design/peons.md) services are task execution engines spawned by Middle Managers. Each Peon runs a separate JVM and is responsible for executing a single task. Peons always run on the same host as the Middle Manager that spawned them.
The Indexer is designed to be easier to configure and deploy compared to the MiddleManager + Peon system and to better enable resource sharing across tasks, which can help streaming ingestion. The Indexer is currently designated [experimental](../development/experimental.md).
Typically, you would deploy one of the following: MiddleManagers, [MiddleManager-less ingestion using Kubernetes](../development/extensions-contrib/k8s-jobs.md), or Indexers. You wouldn't deploy more than one of these options.
Colocating Druid services by server type generally results in better utilization of hardware resources for most clusters.
For very large scale clusters, it can be desirable to split the Druid services such that they run on individual servers to avoid resource contention.
This section describes guidelines and configuration parameters related to service colocation.
### Coordinators and Overlords
The workload on the Coordinator service tends to increase with the number of segments in the cluster. The Overlord's workload also increases based on the number of segments in the cluster, but to a lesser degree than the Coordinator.
In clusters with very high segment counts, it can make sense to separate the Coordinator and Overlord services to provide more resources for the Coordinator's segment balancing workload.
You can run the Coordinator and Overlord services as a single combined service by setting the `druid.coordinator.asOverlord.enabled` property.
For more information, see [Coordinator Operation](../configuration/index.md#coordinator-operation).
With higher levels of ingestion or query load, it can make sense to deploy the Historical and Middle Manager services on separate hosts to to avoid CPU and memory contention.
The Historical service also benefits from having free memory for memory mapped segments, which can be another reason to deploy the Historical and Middle Manager services separately.
- To store all the data you ingest. Segments that get loaded onto Historical services for low latency queries are also kept in deep storage for backup purposes. Additionally, segments that are only in deep storage can be used for [queries from deep storage](../querying/query-from-deep-storage.md).
- As a way to transfer data in the background between Druid services. Druid stores data in files called _segments_.
You can also query directly from deep storage. When you query segments that exist only in deep storage, you trade some performance for the ability to query more of your data without necessarily having to scale your Historical services.
- On disk storage for Historical services need to be able to accommodate the data you want to load onto them to run queries. The data on Historical services should be data you access frequently and need to run low latency queries for.
