Skip to main content

Unleash rawperformance

ObsessionDB. The best ClickHouse® database performance per dollar on the market.

Built for Builders,
Not Babysitters.

We tore apart every node and every byte.
Pure performance is what's left.

No Coordination Layer

No ZooKeeper, no Keeper. No coordination process that can put your replicas into read-only. Just nodes and shared storage.

Stateless Compute

Nodes read from S3 and hold nothing locally. Add one, lose one. The cluster absorbs the change without skipping a query.

One Copy of Your Data

With replicas, you pay for your data three times. Here you pay once, and adding capacity doesn't change that.

Raw ClickHouse SQL

Every INSERT, every DDL, every function in the docs. Not a read-only API, not a managed fork. The full database, nothing removed.

The best performance per dollar.

Choose the plan that fits your workload, with predictable pricing for memory and storage as you grow.

Dev

$15per GB memory
Why we price by memory, not vCPU

For prototyping and development workloads.

Storage

$25/TB · eu-central-1

  • +4-24 GB memory per node, 1-5 nodes
  • +Up to 1 TB storage
  • +Auto-scaling resources with best-effort support
Estimate your monthly cost

Production

$20per GB memory
Why we price by memory, not vCPU

For production analytics at any scale.

Storage

$25/TB · eu-central-1

  • +32-160 GB memory per node, 2-25 nodes
  • +Unlimited storage
  • +Dedicated resources with white glove founder support
Estimate your monthly cost

Enterprise

Custom
Why we price by memory, not vCPU

For large fleets, special compliance, or negotiated capacity commitments.

Storage

$25/TB · eu-central-1

  • +Up to 2 TB memory per node, up to 80 nodes
  • +Unlimited storage
  • +Custom-sized resources with white-glove founder support
  • +Custom SLAs and volume discounts
Estimate your monthly cost

It's just ClickHouse software.

Your existing tools, dashboards, and scripts. One hostname change.

obsessiondb
production ~ $ clickhouse-client \
--host your-cluster.obsessiondb.com \
--password ••••••••
 
Connected to ObsessionDB (ClickHouse 24.12)
 
:) SELECT version()
 
┌─version()─┐
24.12
└───────────┘
 
1 row in set. Elapsed: 0.001 sec.
 
:) SELECT * FROM system.benchmarks FORMAT PrettyCompact

How it Works

Cloud-native engine for ClickHouse workloads with separate compute and storage.

Single Node

ClickHouse
Local Disk
  • Data lives on local disk
  • Limited by single machine resources
  • Scaling requires data migration
  • Single point of failure
  • Must over-provision for peak load

Distributed Tables with Replicas

Distributed Table
Shard 1
R-A1
R-B1
Shard 2
R-A2
R-B2
Shard 3
R-A3
R-B3
ZooKeeper (3+ nodes)
  • Data sharded across nodes
  • Each replica stores full copy of shard data (2× storage)
  • ZooKeeper cluster required for coordination
  • Adding shards requires data redistribution
  • 6+ ClickHouse nodes for basic HA setup

ObsessionDB

Stateless Compute
+
Shared Metadatainstant visibility
Intelligent Cachesub-ms reads
S3 Object Storagesingle copy
  • Compute and storage fully separated
  • Add or remove nodes instantly
  • Single data copy (no replica storage overhead)
  • Shared metadata for instant visibility
  • Intelligent caching + durable S3 storage

Get more out
of your cluster