"What happens to our archive if DMF goes away?" More and more HPE Data Management Framework (DMF) users are asking that question. DMF has kept petabytes of research data on tape for years, moving it off the parallel file system so the file system stays free for active work.
When the future of the software that wrote those tapes is unclear, the archive itself becomes the risk. If you are looking for an HPE DMF alternative, this article covers what DMF does well, the questions to ask before you move, and how NODEUM Data Archiving approaches the same problem with open LTFS tapes.
Over the past months, more and more DMF customers have come to us looking for options. What they tell us is consistent: they are not getting clear answers about the platform's future, and they expect DMF not to be maintained beyond the next two years. For an archive that has to last decades, that is not a comfortable position.
We will be fair to DMF. It is a serious product, and many sites have run it well for a long time. The goal is not to argue that DMF was a bad choice. It is to help you make the next choice with your eyes open.
What HPE DMF does well
DMF is a hierarchical storage management (HSM) system built for large HPC file systems. According to HPE's DMF 7 QuickSpecs, it:
- Migrates files automatically to lower-cost storage when a file system reaches a set threshold, and recalls them when users need them.
- Integrates with HPE XFS, IBM Storage Scale (GPFS) and Lustre, and tracks changes through changelogs instead of full file system scans.
- Writes to a long list of tape libraries and drives, to S3-compatible object storage, and to a power-managed disk tier HPE calls Zero Watt Storage.
- Keeps file metadata in a scalable database, with versioning and multi-site recovery.
That is a lot of capability, and it is why replacing DMF is not a weekend project. Any alternative has to cover the same ground: policy-driven movement, reliable tape handling and recall that users can live with.
Why teams are looking for an HPE DMF alternative
When we talk with storage teams running DMF, the same reasons come up.
Roadmap uncertainty. An archive is a decades-long commitment. If you are not sure the platform will be developed and supported for as long as your data must be kept, you need a plan B before you need it.
Renewal moments. DMF 7 is sold as 1-, 3- or 5-year subscriptions, priced by capacity tier, with extra licences for additional servers. Each renewal is a natural point to ask whether the archive still fits where your storage is going.
Hardware ties. HPE's specifications list specific HPE ProLiant server models for DMF. When those servers reach end of life, the archive refresh and the server refresh become the same project.
Concentrated know-how. At many sites, one or two people really understand the DMF configuration. When they move on, the archive becomes hard to change.
The exit question. Most teams never ask how they would read their tapes without the archive software that wrote them. A roadmap change is the moment that question stops being theoretical.
Five questions to ask before you choose
Whatever you move to, these questions will tell you quickly whether a platform is a long-term home for your archive.
- Can I read my tapes without the vendor's software? If the answer is no, you are not choosing a product, you are choosing a dependency.
- How do users find and recall archived data? Through the file system path they already know, a portal, a catalogue, or a ticket to the storage team?
- Which storage can sit behind the archive? Tape, object storage, cloud, or a mix, with copies placed by policy.
- How is every write verified, and can I prove it later? Regulations and auditors increasingly ask.
- What does migration look like? In phases, with both systems running, or as one risky cut-over weekend?
DMF and NODEUM side by side
NODEUM Data Archiving is an active archive. Instead of hiding tape behind the primary file system, it gives users and applications containers: a container is a virtual file system on which you set placement policies. Behind each container sit secondary pools of type active archive, which can include tape libraries, object storage and cloud. Placement policies decide which data goes where, how many copies exist and when.
| HPE DMF 7 | NODEUM Data Archiving | |
|---|---|---|
| Approach | HSM integrated with the primary file system | Active archive through containers with placement policies; offline archive workflows for data whose tapes can be ejected and kept on shelves |
| File system integration | HPE XFS, IBM Storage Scale, Lustre | Integrates with the file system's HSM mechanism |
| Archive targets | Tape, S3-compatible object storage, Zero Watt disk | Secondary pools: tape libraries, object storage, cloud |
| Tape format | Proprietary format: cartridges are read back through DMF | LTFS: every cartridge is readable with standard LTFS tools, with or without NODEUM |
| Recall | Transparent recall through the file system | Through the container, or from the catalogue for offline tapes, even when they are out of the library |
| Verification | Logical Block Protection on supported tape drives | The archive verifies what was written, for every copy |
The biggest difference is the tape format. With NODEUM, your tapes are still LTFS. If you ever stop using NODEUM, your data is still yours. We explain why that matters in LTFS explained: an open tape format for archives that outlive their software.
Moving off DMF without a big bang
A good migration does not start with a cut-over date. It starts with an inventory.
- Map what you have. How much data sits in DMF, on which tiers, how much of it is ever recalled, and which projects own it.
- Set up the new archive alongside DMF. Create containers and placement policies in NODEUM so that new data goes to the new archive from day one. DMF stops growing.
- Move the data people still use first. Hot and warm archived data moves early, so users see the benefit quickly. Because DMF writes tapes in its own proprietary format, data is recalled through DMF and rewritten to open LTFS tapes, so this is the step where you leave lock-in behind for good.
- Move cold data in the background. Policy-driven, verified transfers run project by project, at a pace your tape drives can sustain. NODEUM Data Archiving has its own workflow engine, built for this kind of large, verified movement to tape.
- Retire DMF when it is empty. Not before.
Run this way, the migration is a series of small, checked steps rather than one irreversible weekend.
Your archive should outlive its software
Tape is still the most economical way to keep research data for decades. What has changed is the expectation that the archive should not lock that data in. Whatever HPE decides for DMF, now is a good time to make sure your next archive keeps your data open, verifiable and easy to recall.
If you are running DMF today and want a second opinion on your options, read our active archive white paper or talk to a NODEUM expert. We will look at your current tiers, your tape estate and what a phased move to NODEUM Data Archiving would involve.