---
title: "VAST Database Overview"
slug: "vast-database-overview-tg"
tags: ["5.5"]
updated: 2026-10-27T23:00:54Z
published: 2026-10-27T23:00:54Z
canonical: "docs.vastdata.com/vast-database-overview-tg"
---

> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vastdata.com/llms.txt
> Use this file to discover all available pages before exploring further.

# VAST Database Overview

The VAST DB is a tabular database that resides on a VAST Cluster, using the extensive storage of the cluster. It provides high performance, bridging the gap between the performance of a data warehouse and the raw scalability of a data lake. It leverages the VAST distributed shared everything (DASE) architecture, together with high-speed low-latency RDMA networking, to provide a database without scaling or performance limitations, and without manageability problems.

It is intended for analytical or very high-volume workloads that can benefit from high-performance and near arbitrary scalability. For example, structured databases that are dozens of petabytes in size, that continuously grow, making the ability to easily load-follow capacity without expensive and disruptive re-sharding tasks; or introducing misalignments between capacity and compute.

The VAST Database is [ACID](https://en.wikipedia.org/wiki/ACID) compliant. All operations are transactions, thus the boundaries of these transactions can be moved at will to cover very complex activities.

In addition, all basic database operations are supported at a granular level. Updates can be across rows or columns and element store mechanics make deletions and updates very low-latency.

VAST Database compute nodes are capable of very efficient low-level capabilities that are exposed to existing query execution engines in the form of push-down operations. When a query execution engine like Trino or Spark generates a query plan, it can offload to the VAST Database parts of the query that the database can handle more efficiently due to its closer proximity to the data.