Proud to present our wide-column-store database

We have launched a wide-column-store database as our first product. It is the basis of our future persistence and artificial-intelligence platform.

Read and write throughput increasing linearly in the number of nodes, maintaining performance.
All nodes are equal (peer-to-peer). There is no master node.
Each node can handle read and write requests independently.
Data is distributed across multiple nodes using consistent distributed hashing.
Nodes can be added or removed without database downtime.
Fast internode streaming is used during scaling operations.

The architecture is designed for continuous availability, as it is masterless, i.e., all nodes are identical, no single points of failure or bottle-necks.
Supports "replication" across multiple data centers for redundancy and disaster recovery.
With the right setup, the database can withstand without downtime node failures or even the outage of an entire data centre. It provides both synchronous and asynchronous replication and repair functionality.

Schema-less design allows dynamic changes to the structure of data, similar to other NoSQL databases). It uses a "column-family" data model inspired by Google (tm) Bigtable.
Allows for "tunable consistency" as the level of consistency (strong or eventual) can be chosen per operation, depending on the use case.
Uses a query language looking similar to SQL but tailored to its architecture.

The persistence mechanisms used in our database are ideal to support RAG (retrieval augmented generation) functionality for LLMs (large language models).
Indexing using SAI (storage-attached indices) provides one of the fastest mechanisms in industry for ANN (approximate nearest neighbor) vector search.
On top, our product will support Java 24 vector processing functionality, supporting SIMD operations, in the near future, to gain additional speed-ups.

Even though it was written in Java, Apache Cassandra (tm) is NOT platform-independent, as it uses native code and hard-coded Linux (tm) path separators. In contrast, our database is platform independent, e.g., you can run it on your MS Windows (tm) laptop or MacBook (tm) instantly.
Apache Cassandra (tm), version 5.0, runs on Java 11 (stable), Java 17 (experimental). Our database runs on Java 24 (only), and is optimized for it.
We support most functionality of Apache Cassandra (tm), version 5.0.4, in particular on the level of its binary protocol. We are committed to be as compatible as possible. However, we deviate from Cassandra via an improved configuration. We added a clear concept for the configuration of the various database directories.

We are committed to "value for money" and a long-term relationship with you.
Our prices are transparent, fair, and competitive.
We think that we can achieve a positive business case even in a minimal setup.
In particular, if you are already using wide-column-store-database services in Public Cloud, there may be an opportunity to save costs. Please do not hesitate to reach out to us, happy to discuss with you!

This is just the start of our
innovative data-persistence and
artificial-intelligence solutions

Name: Zuri-AI GmbH
Telephone: +49(0)15111572555