See the Hammerspace AI Data Platform in Action

Hammerspace AI Data Platform provides a standards-based, high-performance Parallel Global File System. The Dashboard displays real-time performance metrics for your Hammerspace environment. Capacity utilization combines with growth trends for forecasting. Customizable Hammerspace Objectives control data movement across on-premises and cloud infrastructure, ensuring data payloads are positioned for maximum compute performance and data protection. The Alignment pane shows real-time status of data relative to Objectives. Hammerspace AIDP automates objective setting while monitoring for changes to keep data aligned. Assimilation allows the platform to use data from any volume under management, whether on-premises or in the cloud. These volumes can exist across different storage systems. Hammerspace manages storage from any vendor, including cloud resources. Tier 0 resources are Hammerspace orchestrated GPU-local NVMe drives sitting on the PCIe bus with GPUs for AI workloads. Adding a new storage system or cloud instance requires entering storage system details and selecting available volumes. Configure Assimilation Options by setting Source Path from the selected volume and Destination Share and Path for the new Hammerspace location where data becomes visible and available. Read Only Assimilation means the source volume will not be written to by Hammerspace. Existing data can remain in place, or Objectives can be applied to move or create new instances in the background. Additional or Excluded IPs can be configured. The assimilated data becomes accessible through the Destination Share within Hammerspace. The share Dashboard displays performance, capacity and telemetry. Assimilation draws metadata into the Hammerspace Parallel Global File System without moving the actual data. The Hammerspace AIDP Console allows using AI agents to tag files and create collections while updating the AI vector database with embeddings. System Status confirms operational readiness. Natural language prompts tell the AI agent what tasks to perform. Pre-set example prompts enable quick starts. Specify the data path for the agent to work with and define tags for the data. The agent creates a plan and executes, tagging files as instructed and generating vectors for embedding into the vector database. Files can have multiple instances on different volumes. The Hammerspace file system presents it as a single file to connected clients while maintaining identical byte copies on different Hammerspace volumes. A primary instance exists on standard storage while a second instance can be temporarily placed on Tier 0 GPU-local NVMe storage for fastest GPU access. The vector database interface displays the number of vector embeddings created, confirming successful embedding generation from the processed files.