Features

Metadata Indexing

A single persistent source of truth. VAST's DASE architecture enables global algorithms that define how the cluster builds an atomically consistent namespace, enabling super resiliency and linear scale.

Overview

How It Works

Difference

Overview

Inherently Persistent – Why It Matters

Metadata, from basic file names to multiprotocol ACLs and locks, is maintained on shared non-volatile media in VAST enclosures. This allows the mirrored and distributed metadata to serve as a consistent single source of truth. This approach delivers distinct advantages.

Enhanced Resilience: No DRAM write caching means zero volatility in the data path and no need to rebuild data when Servers fail. Data structures are always atomic, eliminating the need for cache coherency.

Linear scale: Without the need for east-west cluster traffic every Server and Enclosure adds a proportionately linear amount of performance to the VAST Cluster.

How It Works

Transactionally Consistent

VAST Element Store’s namespace metadata can be thought of in some ways as a database against which the system makes queries to locate pieces of data by file or object name, snapshot time and other metadata attributes. That database metaphor also extends to how the VAST Element Store uses transactional semantics to ensure that the VAST Element Store, like a relational database, is fully ACID (Atomic, Consistent, Isolated, Durable).

To remain consistent, the VAST Element Store ensures that each transaction is atomic. To achieve this, each storage transaction is either applied to the metadata (and all of its mirrors) in its entirety or not applied to the metadata at all, even when a single transaction updates many metadata objects). Since every transaction to the element store is atomically consistent, so is the element store metadata as a whole which eliminates the need for classic file system check tools like the dreaded fsck and journal replays so systems can be instantaneously functional upon power cycle events.

Difference

Eliminate the complexity and limitations of shared-nothing systems with DASE.

Scalable to exabytes, this new architecture delivers linear performance scale by eliminating communication across storage CPUs, server rebuilds, and interdependencies that increase geometrically with cluster size.