The ScoutAM platform utilizes the open-source Scale Out Filesystem (ScoutFS) to maintain and manage metadata. ScoutFS was designed to manage tens to hundreds of billions of files and objects in a single namespace.  ScoutFS is a multi-node, clustered metadata management system. In the ScoutFS architecture, there is no central metadata server.  All of the nodes in a ScoutAM cluster (or a subset of the nodes) may be utilized by the ScoutFS filesystem allowing the combined resources of the cluster to be leveraged for scalable metadata services.

ScoutAM is the only mass storage platform built on a converged metadata architecture, meaning that metadata is stored within the application stack.  Other products utilize a diverged model wherein metadata is handed off to a separate third-party database tool like Cassandra, or DB2 for storage and processing. Many of the legacy mass storage systems refer to their metadata collection as “the database”, reflecting the fact that the metadata is managed by a separate tool.

In addition to adding significant complexity to the application, external databases must be populated by transferring or handing off metadata from an underlying filesystem source.  The hand-off introduces latency and is error prone even under optimal operating conditions.  When errors occur, the ingested data stream must be interrupted and replayed.  If the database server is unable to keep up with the pace of incoming metadata information, then the database create rate establishes a fixed upper boundary on performance for the entire system. By contrast, ScoutAM utilizes ScoutFS to manage metadata internally with its own highly optimized data structures. This converged architecture is simpler, faster, more reliable, and more scalable.