From the NetApp edition of November 2, 2026
Supercomputing News: PEAK:AIO Lattice pNFS metadata server, built with Los Alamos and CMU, is an early release with vendor-only benchmarks and an unclear ONTAP integration path
Supercomputing News (SCN staff) · Independent HPC trade reporting · United States / UK
- Publisher
- Supercomputing News
- Original publication
- 28 September 2026
- Source retrieved
- 2026-11-03
What happened
SCN reviewed the deal announcement and Lattice materials. Lattice is an open-source NFSv4.2 metadata server PEAK:AIO developed with Los Alamos National Laboratory and Carnegie Mellon University. SCN reports that PEAK:AIO and LANL launch figures claimed 40 GB/s versus 3-7 GB/s for standard Linux NFS on the same LANL servers, gains from 70 to 400 GB/s, and up to 10x on mdtest, but some figures lack baselines and SCN found no measurement of application throughput or GPU utilization. The project README calls Lattice an early release not recommended for production. SCN notes ONTAP requires NFS referrals disabled to enable pNFS, while Lattice depends on referrals, so integration depends on a design NetApp has not described.
Why it matters
Newly relevant because the acquisition closed on 2 November. It gives research universities and national-lab-adjacent centers an independent check on NetApp's claims: the underlying technology came from public research labs but has no independent production benchmarks, so campus HPC buyers should wait for measured results.
Evidence and measured results
Opened the full article (published 28 Sep 2026). It is independent trade analysis based on vendor and project documents, not its own testing; benchmark numbers are PEAK:AIO and LANL claims reported by SCN.
Limitations and uncertainty
Predates the edition window; written before close. Benchmarks are vendor- and lab-reported, some without baselines. SCN's ONTAP conflict reading is its own interpretation.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-02; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Research centers hear large parallel-NFS performance claims but lack independent evidence for their workloads.
- Stakeholders
- Research computing director
- HPC storage engineers
- Vice president for research
- Information security officer
- Discovery questions
- Which workloads are bound by metadata operations?
- Do you have a test cluster for storage evaluations?
- What evidence do you require before production storage changes?
- Value hypothesis
- A workload-based evaluation can show whether parallel metadata helps before any purchase.
- Potential engagement
- A bounded non-production evaluation plan for one research cluster.
- Unsupported claims
- 40 GB/s or 10x gains at your site
- Production readiness
- Seamless ONTAP integration
Pre-sales engineering
Role takeaway
- Fit
- Metadata-heavy research AI and simulation workloads on NFS.
- Architecture and integration
- Multiple metadata servers on shared RonDB with NFSv4 referral routing; unpatched Linux knfsd data servers; ONTAP fit unresolved.
- Prerequisites
- Test cluster
- Representative workloads
- Baseline NFS measurements
- Constraints
- Early release, not production
- Vendor-reported benchmarks
- ONTAP referral conflict unresolved
- Security
- Run independent security review before any production use, as the README advises.
- Proof of value
- Measure application throughput versus NFS baseline
- Measure GPU utilization changes
- Complete security review
Delivery
Role takeaway
- Work
- Run a non-production Lattice evaluation with baselines and security review.
- Dependencies
- Test hardware
- Staff time
- Commercial support clarity from NetApp
- Ownership
- Research computing director.
- Skills and adoption
- HPC engineers learn pNFS Flex Files and metadata server operations.
- Governance checkpoints
- No production use without validation
- Security review sign-off
- Acceptance criteria
- Measured workload throughput gain over baseline (proposal)
- No unresolved high-severity security findings (proposal)
- Risks
- Immature software
- Unclear post-acquisition support
- Integration uncertainty
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Lattice uses multiple metadata servers on a shared RonDB store with clients routed by NFSv4 referrals; conflicts with ONTAP's pNFS-referral rule may need reconciling.
Governance
Who approves, reviews and stays accountable for outcomes?
Procurement should require independent benchmarks before relying on parallel-metadata claims.
Security and privacy
What data, permissions and controls need testing?
Project README calls for independent security review before production use.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
None stated.
Procurement
What should contracts, pricing and exit terms secure?
Open-source Lattice code is MIT-licensed with one GPL-2.0 file, per related reporting; commercial packaging after acquisition is unknown.
Operating model
Which teams own the service once it runs?
Research computing centers can evaluate Lattice in test environments independently of NetApp's roadmap.
Publication history
- 2026-11-02NetApp · Issue 043 resources
Stable resource ID: scn-peak-aio-lattice-pnfs-early-release-no-independent-benchmarks-20260928