Apache Pinot vs ClickHouse

Compare and contrast Apache Pinot and ClickHouse by architecture, ingestion, queries, performance, and scalability.

Compare Apache Pinot to Rockset here

Compare ClickHouse to Rockset here

Apache Pinot vs ClickHouse Architecture

Architecture

Apache Pinot

ClickHouse

Deployment model

PaaS or self managed

• Self-managed on-premises or on cloud infrastructure • Several managed cloud services available

Use of storage hierarchy

Hot storage plus Deep Store for backup and restore operations

• Designed to use hard disk drives for storage • Can also use SSD if available

Isolation of ingest and query

Separation of compute and storage

No, although StarTree separates compute and cloud storage, which is an important consideration for those considering StarTree vs. open-source Pinot

No - although ClickHouse Cloud decouples compute and cloud storage

Isolation for multiple applications

Full isolation with replication

Pinot is a real-time distributed OLAP datastore that ingests both batch and streaming data. It has a distributed systems architecture that scales both horizontally and vertically, but unlike alternative OLAP databases, it does not decouple storage and compute. It supports both self-managed and PaaS options.

ClickHouse is open source and can be deployed anywhere. Several vendors such as ClickHouse, the company, and Tinybird also offer cloud versions. Compute and storage are tightly coupled, although ClickHouse Cloud was rearchitected to decouple compute and storage. ClickHouse Cloud pricing is based on compute and storage usage.

Apache Pinot vs ClickHouse Ingestion

Ingestion

Apache Pinot

ClickHouse

Data sources

Streaming • AWS Kinesis • Apache Kafka Batch • Cloud Storage • PostgreSQL • MySQL • Snowflake • Google BigQuery File upload Write API

• Core integrations for ingestion from Kafka, S3, Google Cloud Storage • Other partner and community integrations available

Semi structured data

Support for JSON column types, but JSON must first be converted to a string

• JSON Object type for handling nested JSON • Automatically infers schema from a subset of rows

Transformations and rollups

Yes

Yes - several storage engines can pre-aggregate data

Pinot supports high-performance ingest from streaming data sources. Each table is either offline or real time. Real-time tables have a smaller retention period and scale based on ingestion rate while offline tables have a larger retention period and scale based on the amount of data. In order to persistently store the generated segments that make up a table, you will need to change controller and server configs to add deep storage.

ClickHouse has core integrations from common sources such as Kafka and S3. It recently introduced greater ability to handle semi-structured data using the JSON Object type and automatic schema inference.

Apache Pinot vs ClickHouse Performance

Performance

Apache Pinot

ClickHouse

Updates

• By default, all data in Pinot is immutable • Upserts only supported for streaming ingest • No support for upserts on data using the star tree index • No support for upserts on out-of-order events

• Writes to immutable files • Updates rewrite and merge data files asynchronously • Frequent updates are not recommended due to potential for large rewrites

Indexing

• Manually configured • Forward index, inverted Index, Star-Tree index (columnar), bloom filter, range index, search index, JSON index, geospatial index, timestamp index

• Primary indexes use sparse indexing on data ordered by primary key • Secondary data skipping indexes

Query latency

50-1000ms queries on 100s of TB

Sub-100ms to seconds, optimized for large-scale aggregations

Storage format

Pinot stores data in a columnar format and adds additional indexes to perform fast filtering, aggregation and group by

• Column-oriented • Heavily compressed to minimize storage footprint

Streaming ingest

1-2 second ingest for streaming data

Recommends inserting in batches of >1000 rows and <1 insert per second

Like its competitors, Pinot can achieve sub-second query latency at high concurrency. However, this level of performance requires tuning, management, and deep expertise. Compared with the open-source version, the PaaS versions of Pinot address some of these issues, but similarly require expertise while making tradeoffs affecting query performance.

ClickHouse leverages column orientation and heavy compression for better performance on analytics workloads. It also uses indexing to accelerate queries as well. While ClickHouse use cases often involve streaming data from Kafka, batching data is recommended for efficient ingestion.

Apache Pinot vs ClickHouse Queries

Queries

Apache Pinot

ClickHouse

Joins

SQL joins limited to inner join, left-outer, and semi-join

Yes

Query language

SQL

Developer tooling

• API for querying SQL via POST command • Clients for JDBC, Java, Python, and Golang • Integrations with Trino and Presto

• API for querying SQL via POST command • Python, Java, Node.js and Go language clients

Visualization tools

Integrations with Tableau and Superset

Integrations with Metabase, Superset, Grafana, Tableau, Deepnote and Rocket BI

In Pinot, SQL-like queries are received by brokers and scatter the request between real-time and offline servers. The two tables then process requests, send results back to the broker, and responds with the result. Joins are limited, as is support for UDFs and subqueries, making Pinot more or less useful depending on the use case.

ClickHouse uses SQL for querying, with support for SQL joins. ClickHouse integrates with some common tools for visual analytics, including Superset, Grafana and Tableau.

Apache Pinot vs ClickHouse Scalability

Scalability

Apache Pinot

ClickHouse

Vertical scaling

All 4 pinot node types can be resized manually

Scale up single-node ClickHouse for vertical scaling

Horizontal scaling

• Users can add additional Pinot nodes to scale horizontally • Rebalancing is manual

• Compute and storage scaled in lockstep • Data rebalanced to populate newly added nodes • Cloud offerings automate some of the scaling and rebalancing effort

Pinot allows for vertical scaling by increasing CPU and memory for each node as well as horizontal scaling by adding additional nodes. Capacity planning is a time-consuming, iterative, and manual task. It involves load testing and tuning across multiple vectors including read QPS, write QPS, number of streaming partitions, daily data size, retention period, types of workloads, number and type of segments, and much more.

ClickHouse can be used in both single-node and distributed modes. Tight coupling of compute and storage and the need to rebalance data make scaling out more complex, but cloud versions of ClickHouse help automate this process.