![]()
Most enterprise unstructured data, typically about 80%, is cold, yet it still sits on expensive primary storage. Storage teams know they should tier and archive more of it. The latest Komprise Intelligent Data Management release delivers updates to improve cold data tiering in addition to enhancements for AI data governance and search and migrations.
Komprise customers should read the release notes (login required) and/or connect with their account team to learn the specifics. Here’s a high-level summary of what’s new:
- Tiering at scale: Plans with many targets need far less Observer memory, so you avoid costly VM upgrades as you tier more data.
- Governed data for AI: AI assistants query Komprise directly through the new Universal File MCP Server.
- Faster deployments: Analysis and migration projects start without joining Observers to Active Directory.
Many of these updates come straight from customer and partner requests. Here is what’s new and why it matters to the teams who run your IT infrastructure.
Tier More Data with Less Effort
Komprise Transparent Move Technology (TMT) typically cuts storage costs by 70% or more by automatically moving cold data off expensive primary storage while users and applications keep the same access and there is no rehydration penalty. The latest release removes two of the biggest reasons teams tier less than they could: infrastructure overhead and user anxiety.
Target virtualization reduces memory needs
Organizations that tier to many targets used to add memory to their Observers to keep up, because each target needed its own process. With server memory prices up sharply since late 2025, that overhead has become a real line item. Target virtualization consolidates targets behind a single process, so Observers use far less memory. You can add targets without resizing your Observer VMs or paying for larger cloud instances.
Distributed NFS access for better performance
When hundreds of clients read tiered data over NFS, every request used to flow through a single lead Observer. Distributed NFS access spreads those requests across the Observer grid, so performance holds up under heavy parallel load and no single Observer becomes a bottleneck. This option is available for clients using NFS v4.1 or v4.2.
Bulk recall allows users to find the exact files they need
The most common objection to tiering is simple: “What if users need it back?” Previously, recall worked at the share or directory level, and only admins could run it. Bulk recall by query changes that. Run a Deep Analytics query, then recall only the files it finds. Additionally, a new Deep Analytics + Recall role lets admins delegate recall to data owners for the specific shares they own without filing a ticket.
Ringfence data with updated Confine functionality
Confine folder permissions are now configurable. Keep confined data visible only to admins or keep the original permissions intact so users can browse the confined data. Komprise also removes empty folders left behind after a move, so reporting scripts and ingestion pipelines stay clean.
Deliver Governed Data to AI and Analytics
Transparent File Tables delivers query-ready file data to the lakehouse. Transparent File Tables, now generally available, exports the complete results of a Deep Analytics query from the Global Metadatabase as CSV or Apache Iceberg-compatible Parquet, ready for Snowflake, Databricks, Microsoft Fabric and other data lakehouse ingestion. One early customer exported metadata for about 40 million files in under four hours. Exports go to Amazon S3, Azure Blob or NAS.
The Komprise Universal File MCP server bridges the gap between users who want answers about file data without waiting on IT, and IT needs to control access..
-
- Users connect their GenAI tools (ChatGPT, Copilot, Claude, Cursor, etc) to Komprise and ask questions about their data in plain language.
- Each question becomes a direct API call to Komprise, so the data behind every answer comes straight from the Global Metadatabase, not from the model’s guesswork.
- Every user signs in with their own credentials and the Universal File MCP Server honors the same role and share-based permissions as the Komprise UI. Deep Analytics users can now use the MCP Server, not just admins. The MCP Server is available through the Komprise Early Access Program. Contact your account team or email eap@komprise.com to participate.
A managed tag library gives control and standardization to tags
Tags now drive Smart Data Workflows, KAPPA data services, exports and archiving policies, so consistency matters. The new tag library lets admins create, rename and delete tags centrally, including bulk creation from a CSV. Renaming a tag updates every file that carries it. Users apply tags from the approved library instead of typing new ones, which keeps typos out of the policies that depend on them.
ACL reporting proves who can access what data
Security and compliance teams need evidence of who can access which files. ACL reporting for SMB and dual-protocol shares generates a path-level report of every access control entry on every file and directory, with SIDs resolved to user and group names. It reads metadata only and never changes data, which makes it a practical starting point for audits and access reviews.
Deploy Faster and Migrate with Full Visibility
There are many other updates across the Komprise Intelligent Data Management platform, including:
- Analysis, migration and confine-only projects run without joining Observers to Active Directory. Joining Observers to the domain has long been the most common reason a deployment or proof of concept slows down as it triggers security reviews and permissions troubleshooting.
- See every migration error, not a sample in Elastic Data Migration. The full list of migration errors appears in the UI, paginated, with a one-click download of the complete set.
- IBM Storage Scale (GPFS) is now a standard data store, accessed over standard file protocols, opening tiering and analytics for HPC, research and life sciences environments.
- Google Cloud Storage at scale: Discover every bucket in a project and run bulk actions across hundreds of buckets at once, instead of adding them one at a time.
- Spectra Logic BlackPearl: Observers now apply connection defaults automatically, so tiering to tape-backed object storage stays within device limits without manual tuning.
Get the Latest Release
Komprise 6.6 is the recommended release for customers and new deployments and it is required to use the Komprise Universal File MCP server.
Customers should review the release notes and talk to your Komprise team about upgrading and joining the MCP Early Access Program.
Also be sure to register for the Dec. 3rd Office Hours webinar with Komprise Field CTO Benjamin Henry: Komprise Universal File MCP: What’s New and What’s Next: komprise.com/office-hours


