No distributed HTAP database is faster

One engine, scaling across machines, serving transactions and analytics from the same storage. Against every other database that claims that, on the same 16 vCPU AWS instance and the same query set, ScramDB is first. Not on one measure. On every one of them.

clear of the next distributed HTAP database.

Measured 5 September 2026. A snapshot of that day, and the engine has not stopped getting faster since.

#1of 10 distributed HTAP databases

#1from a cold start

#1on storage footprint

377xfaster than PostgreSQL on the same queries

Time to answer the query set with the data warm. This is the number a dashboard, a report or an agent actually waits on.

Databases that scale across machines AND serve transactions and analytics from their own storage. This is the category ScramDB competes in, and it leads every measure of it.

1ScramDBScramDBx1.00

2SingleStoretunedx4.03

3CrateDB (tuned)tunedx12.71

4CrateDBx15.13

5OceanBase (row store)x65.05

6Citusx86.59

7TiDB (TiFlash only)x360.97

8TiDB (TiKV only)x366.48

9CockroachDBx986.89

10YugabyteDBx4888.45

How to read this, as of 5 September 2026. Every figure is a measurement on the same instance type: rival figures are those systems’ own published results, and the ScramDB row is our own run of the same query set, on a single node in its stock configuration with the GPU off and nothing tuned for the test. Lower is better on every metric, and each one is shown relative to the best result in the field. These are the numbers as of that date, not a permanent scoreboard: the engine is tuned continuously and it has got faster through every round of this work, so a later run should read better than this one.

What a table cannot show

First in the category that matters

Widen the field to every engine measured and a handful sit above ScramDB on raw query speed. Look at what they are: research prototypes, engines embedded inside another product, analytics plugins their own projects label alpha, and query engines that own no storage at all. Not one of them holds your transactions, and not one of them scales across machines while doing it.

Among databases that actually do both, ScramDB is first, and first by more than an order of magnitude. It gets there with serializable isolation, the full PostgreSQL surface, instant branching, a semantic layer agents connect to directly, and clustering that spans regions on the same binary.

See ScramDB against your database

Run it yourself in one line

One static binary, no dependencies. It picks its own config, listens on the postgres port 5432 and runs in the background.

$curl -fsSL https://scramdb.com/install | bash

Quick start · How the engine works · Talk to us