ClickHouse

ClickHouse is an open-source, column-oriented SQL database built for online analytical processing (OLAP): scanning and aggregating billions of rows in milliseconds. It is not a Solana protocol and has no token. It appears in the Solana ecosystem as data infrastructure. CryptoHouse, built with Goldsky, gives anyone free SQL access to Solana blocks, transactions and token transfers at crypto.clickhouse.com. Triton One's open-source Superbank and RPC 2.0 historical layer store Solana's full ledger in ClickHouse. ClickHouse began inside Yandex in 2009, was open-sourced in 2016 and became an independent company, ClickHouse, Inc., in 2021. As of September 2026 it has more than 4,000 ClickHouse Cloud customers and a $15 billion valuation.

How ClickHouse works: a columnar OLAP database

Row-oriented databases such as PostgreSQL and MySQL are built for transactions: reading and writing individual records. ClickHouse stores data column by column instead. A query that touches five columns of a 200-column table reads only those five from disk. That layout compresses well and cuts I/O, which is why ClickHouse says column-oriented systems are "at least 100x faster in processing most queries" in OLAP workloads (clickhouse.com).

• Vectorized execution: the engine processes data in blocks rather than row by row, using CPU SIMD instructions and cache locality.
• Standard SQL: GROUP BY, JOINs, window functions and subqueries, so analysts don't need a proprietary query language. The 26.8 release (August 2026) added a |> pipe operator for writing queries as a sequence of transformations (September 2026 newsletter).
• Replication: asynchronous multi-master replication across nodes for availability.
• Materialized views: pre-computed aggregates that update as data arrives. CryptoHouse leans on these for fast Solana queries.

The trade-off is the flip side of that design. ClickHouse uses a sparse primary index, so it is fast at time-ordered scans and aggregations and comparatively slow at uniformly random point lookups. That distinction drives the Helius decision covered below.

CryptoHouse: free Solana SQL analytics on ClickHouse

CryptoHouse is a free, public blockchain analytics service from ClickHouse and Goldsky, launched on August 7, 2024 (announcement) and presented at Breakpoint 2024. Goldsky converts chain data into structured tables and streams them in through its Mirror pipelines. The Solana Foundation worked with ClickHouse on the materialized views used to speed up common queries.

Per Goldsky's docs, CryptoHouse covers these Solana datasets, with similar datasets for Ethereum and Base:

• blocks: block-level metadata
• transactions: full transaction history
• token_transfers: SPL token movements
• block_rewards: validator and staking rewards
• accounts: account state
• tokens: SPL token metadata

At launch the Solana transactions table alone held 79.34 TiB compressed (468.91 TiB uncompressed), with a non-voting transactions table of 17.89 TiB compressed. Most of Solana's raw transaction volume is validator vote traffic, so for user-activity questions the non-voting table is far cheaper to scan.

How to query Solana data in CryptoHouse

1. Open crypto.clickhouse.com and write SQL in the browser. No account, credit card or compute setup is needed, and results come with basic charting.
2. To connect a client or script, use the standard ClickHouse client with --host crypto-clickhouse.clickhouse.com --secure --user crypto and an empty password (CryptoHouse repo).
3. Start from the example queries and materialized views in the CryptoHouse GitHub repo, where you can also raise issues and request new views.

Free access comes with quotas. Each query can scan at most 10 billion rows and run for 60 seconds, and each user gets 60 queries per hour (announcement). Full-history scans of raw transactions will hit those limits, so filter by time range or use the pre-built views.

ClickHouse in Solana infrastructure: Triton Superbank and RPC 2.0

The biggest Solana use of ClickHouse in 2026 is in RPC history. Triton One's Superbank replaces Google Bigtable with ClickHouse as the store for Solana's ledger since genesis: "400+ TB of history, 400+ billion transactions, and trillions of rows" (Triton blog). Triton reports these speedups for the methods that were live at release:

Superbank is open source under the AGPL. It is also the historical module of Triton's RPC 2.0, announced with the Solana Foundation in May 2026. In RPC 2.0, ClickHouse "derives its own query-optimised views from the raw data, each one physically sorted to match how Solana apps actually read history" (Triton blog). We covered the release in Triton One open-sources SuperBank.

Outside the RPC layer, ClickHouse is a common backend for Solana indexers, explorers and DeFi dashboards. Community projects such as a Geyser plugin for ClickHouse stream account updates straight into it.

ClickHouse vs RocksDB for Solana archival data

The verdict depends on the access pattern. For time-ordered analytics and aggregation ClickHouse is the stronger fit. For lookups by transaction signature at very large scale, a key-value store can win.

Helius hSOL$125.58+2.6% ran its Solana archive on ClickHouse, then moved "over 300 terabytes" of it to RocksDB in 2026 (Helius blog). Its reasoning: Solana signatures are 64-byte random identifiers with no time locality. Under ClickHouse's sparse index, "a single signature lookup could touch 20 granules", roughly 10,000 rows and about 60 disk I/Os, and batches of 1,000 transactions approached a full second. After the move, p95 latency for getTransactionsForAddress fell from 350 ms to 30 ms, and compressed storage shrank from about 330 TB to about 190 TB.

Triton went the other way, choosing ClickHouse for the same ledger and sorting its views around how apps read history. Two experienced Solana RPC teams reached opposite conclusions. The lesson is that schema and access-pattern design matter more than the engine's name.

ClickHouse Cloud pricing and deployment options

• Open source: self-hosted under the Apache 2.0 licence, free apart from your own infrastructure.
• ClickHouse Cloud: a managed service on AWS, Google Cloud and Azure, with usage-based billing starting at $50 a month (clickhouse.com). It handles provisioning, scaling, backups and upgrades. On-Demand Compute, which adds workers to individual heavy queries, entered private preview in September 2026.
• Embedded and local: chDB runs ClickHouse in-process, and ClickHouse Local queries files without a server.
• Managed Postgres: a Postgres service, announced in January 2026, that syncs transactional data into ClickHouse. It reached Google Cloud in private preview in September 2026.

Who should use ClickHouse for Solana data

• Analysts and researchers who want free, ad hoc SQL over Solana history without running infrastructure. CryptoHouse fits as long as your queries stay inside the 10-billion-row, 60-second and 60-per-hour limits.
• Teams building indexers, dashboards or explorers that aggregate large volumes of time-ordered on-chain data. Self-hosted ClickHouse or ClickHouse Cloud is a common backend, fed by Geyser plugins, Goldsky pipelines or similar.
• RPC operators and archival services should benchmark against their own lookup pattern first. As the Helius and Triton cases show, signature-keyed point lookups can favour a key-value store unless the ClickHouse schema is built around them.

It is a poor fit for people who want ready-made Solana charts without writing SQL, or for transactional application state that needs frequent single-row updates.

ClickHouse risks and limitations

• No on-chain footprint to audit: ClickHouse has no Solana programs, token or custody of user funds, so smart-contract risk does not apply. The risks are operational.
• CryptoHouse is a free community service: it publishes no uptime commitment, and its data depends on Goldsky's ingestion pipeline. Quotas cap heavy queries, so don't build production systems on it.
• Random-key lookups: the sparse index makes signature or account point lookups expensive at petabyte scale unless the schema and sort keys are designed for them, as the Helius migration shows.
• Operational complexity when self-hosting: schema design, sort keys, merges and replication need real database expertise. The managed Cloud service trades that work for vendor dependence and usage-based costs.
• Data freshness: Solana data in ClickHouse is only as fresh and complete as the pipeline that feeds it. Results can lag the chain or differ from an RPC node's view.

ClickHouse alternatives for Solana analytics

• Dune and Flipside: hosted blockchain SQL platforms with curated, decoded Solana tables and community dashboards. They are friendlier for non-engineers. CryptoHouse is better for raw, fast SQL over Goldsky's tables.
• Google BigQuery: hosts public crypto datasets and suits teams already on Google Cloud, but billing is per bytes scanned, which gets expensive on Solana-scale tables.
• RocksDB and other key-value stores: better for signature-keyed archival lookups, as Helius chose. They are not an analytics engine.
• Snowflake and other cloud warehouses: general-purpose warehouses that ClickHouse competes with for analytics workloads.

Does ClickHouse have a token?

No. ClickHouse has no cryptocurrency, staking or on-chain governance. Any token marketed under the ClickHouse name is not affiliated with ClickHouse, Inc. The company earns revenue from ClickHouse Cloud subscriptions.

ClickHouse company: funding, valuation and revenue

Alexey Milovidov led ClickHouse's original development at Yandex from 2009, where it powered the Yandex.Metrica web analytics platform. He is CTO of ClickHouse, Inc., which lists offices in the San Francisco Bay Area and Amsterdam.

• Funding: a $50 million Series A (2021, Index Ventures and Benchmark), a $250 million Series B (October 2021, led by Coatue and Altimeter), a $350 million Series C (May 2025, led by Khosla Ventures) and a $400 million Series D at a $15 billion valuation (January 2026, led by Dragoneer). Bessemer, GIC, Index Ventures, Khosla Ventures, Lightspeed, T. Rowe Price-advised accounts and WCM also took part in the Series D (Yahoo Finance). The four rounds total more than $1 billion.
• Revenue and scale: in September 2026 the company said it had passed $350 million in run-rate revenue, up from about $200 million at the end of 2025. It reported more than 4,000 customers, including DoorDash, Ramp, Meta, Tesla, Cisco and Visa, and nearly 800 employees in 27 countries (ClickHouse).
• IPO signals: on September 22, 2026, ClickHouse added Mike Scarpelli, former CFO of Snowflake and ServiceNow, to its board as an independent director and audit committee chair. He took both of those companies public. ClickHouse has not announced an IPO.
• Acquisitions: these include PeerDB (Postgres change data capture), HyperDX (observability, now part of ClickStack), LibreChat (an open-source AI chat interface) and Langfuse (LLM observability, announced with the Series D).