Where LStore comes from.

Unique Checksum publishes this provenance because honesty about where LStore comes from is part of representing it well. The firm did not invent LStore, and the value of the platform to an institution rests substantially on the fact that LStore is the product of a long, serious research-and-engineering effort by others — not a product Unique Checksum could have built and is now selling. Getting the credit right is both a matter of accuracy and a matter of the platform’s credibility: the lineage is the evidence that LStore is what it claims to be.

The lineage has three movements. First, the foundational research — logistical networking and the Internet Backplane Protocol — that established the ideas LStore is built on. Second, the engineering at Vanderbilt’s ACCRE that turned those ideas into a production file system over more than two decades. Third, the present chapter, in which Unique Checksum carries the platform to institutions outside Vanderbilt under contract to ACCRE. Each movement is the work of different people, and this page keeps the credit where it belongs at each stage.

The research originated at the University of Tennessee’s Logistical Computing and Internetworking laboratory, where the Internet Backplane Protocol was designed as a mechanism for managing and using remote storage in large-scale distributed systems. The IBP research effort was led by Dr. Micah Beck, who founded and directs the laboratory and built the broader logistical-networking research program around the protocol, together with collaborators including Terry Moore. Dr. James S. Plank’s contribution that LStore depends on today is the separate Jerasure erasure-coding library — an open-source Reed-Solomon implementation that protects data in LStore and is covered on Reliability & Integrity. The earliest IBP technical reports date to the late 1990s; the protocol is described in the laboratory’s foundational 1999 technical report on the Internet Backplane Protocol and in the subsequent IEEE publication.

Among the researchers who built that foundation was Scott Atchley, then at the University of Tennessee, a significant contributor to the early IBP and logistical-networking work — including the fault-tolerance design of the network storage stack and the logistical-networking testbed through which IBP, the L-Bone depot directory, and the exnode-based storage tooling were demonstrated at scale. He is today the Chief Technology Officer of the National Center for Computational Sciences and the Oak Ridge Leadership Computing Facility at Oak Ridge National Laboratory — one of the senior technical leaders in U.S. Department of Energy scientific computing.

Two ideas from that research are load-bearing in LStore to this day. The first is IBP’s treatment of storage as a leased, time-limited, capability-addressed resource rather than permanent local property — the deliberately fluid foundation that the architecture page describes as Layer 1. The second is the exnode: the research introduced the exnode as, in its authors’ own framing, an inode for network-distributed files — a metadata object describing how a file’s data is spread across distributed storage. The exnode that the exnode page describes as the heart of LStore’s per-file policy model descends directly from this research. The vocabulary of LStore — depots, exnodes, logistical I/O — is the vocabulary of logistical networking, carried forward.

This is the lineage that gives LStore its conceptual depth. The platform is not an assemblage of recent engineering convenience; it is the production maturation of a research program that thought carefully, more than two decades ago, about what it would mean to treat storage as networked infrastructure. The ideas were proven in research first, and the engineering that followed built on a foundation that had already been reasoned through.

• Late 1990s–2000s

  Logistical networking research

  The Internet Backplane Protocol and the exnode are developed as research at the University of Tennessee’s Logistical Computing and Internetworking laboratory, establishing storage-as-network-resource as a workable architecture.

• 2006–2011

  REDDnet

  The Research and Education Data Depot Network — an NSF-funded wide-area storage project centered at Vanderbilt — deploys geographically distributed depots using logistical-networking middleware. LStore began as the file system layer of this effort, with collaborators across many sites and storage centrally managed at Vanderbilt as a shared global file system.

• 2011 onward

  CMS Tier 2 production

  Vanderbilt becomes a Tier 2 institution for the CMS experiment at CERN’s Large Hadron Collider, with the data stored in LStore. Production high-energy-physics data at multi-petabyte scale validates the platform under one of the most demanding data workloads in science.

• Two decades on

  Mature production platform

  LStore runs as ACCRE’s multi-petabyte research storage platform, developed in-house to meet needs that off-the-shelf storage could not, and refined continuously against real production demands across the university’s research community.

The decision to build LStore in-house was a response to a real constraint: ACCRE’s researchers had storage needs that commercial off-the-shelf systems did not meet, and the logistical-networking foundation offered a path to meeting them that no purchased product did. That origin matters to the platform’s character. LStore was not designed to be sold; it was designed to solve the actual problems of a working research-computing center, refined over two decades against the unforgiving feedback of production use. The qualities that make it valuable — its tolerance for heterogeneous hardware, its per-file placement policy, its erasure-coded durability — are the qualities a real center needed, not features chosen for a product sheet.

REDDnet itself was, in the words of the project’s public description, a wide-area storage network of geographically distributed depots using Internet data striping with fault-tolerant design. The fault tolerance, the distribution, the depot model — the architecture the rest of this site describes — were present from the start of the Vanderbilt effort, because they were the point of it.

The project began in Dr. Micah Beck’s research group at the University of Tennessee, in the Logistical Computing and Internetworking laboratory that Dr. Beck founded and directs. The foundational work on the Internet Backplane Protocol and the exnode — the ideas LStore is built on — was done there, with staff time paid largely by external research grants flowing through the laboratory, principally from the National Science Foundation. Two people who would later carry the project forward at ACCRE were already in or near that group: Mr. Hunter Hagewood was a student in Dr. Beck’s group and worked on the project in that period, and Dr. Alan Tackett would shortly join the broader logistical-networking effort through REDDnet.

From there the project moved to REDDnet, the Research and Education Data Depot Network, launched in 2006 with funding from the National Science Foundation (grant PHY-0619847) and the Vanderbilt Center for the Americas. REDDnet carried the logistical-networking work out of pure research and into a wide-area, multi-institution storage testbed centered at Vanderbilt. Dr. Tackett took up the project within REDDnet and has been the continuous technical through-line on its development from that point to today — the rare case of a complex software platform with a single lead engineer present across more than two decades of its life.

The transition into ACCRE proper — the move from a research-network project to a production file system inside Vanderbilt’s research-computing center — came through Dr. Paul Sheldon, Professor of Physics and Astronomy (now Emeritus), co-founder of ACCRE in 2003 and its longtime faculty chair. Dr. Sheldon’s CMS particle-physics work at the Large Hadron Collider needed a storage platform that could carry petabyte-scale experimental data reliably and at low cost. He brought LStore into ACCRE as the storage layer for that work, and the CMS Tier 2 production use that the platform has carried since 2011 dates from that decision. The funding model from that point forward has been the combination that supports it today: federal grant funding through the CMS Heavy Ion (CMS-HI) program at the Department of Energy, together with institutional contributions from Vanderbilt University. Dr. Tackett continued as the lead developer; Mr. Hagewood, after a career in the private sector, returned to ACCRE and is now part of the program again as its Director of Operations and Interim Executive Director.

What Vanderbilt provides is real and ongoing, and worth naming clearly. ACCRE is the platform’s institutional home: the production environment in which LStore has been hardened against more than two decades of real workloads; the long-term operational staff that keeps it running at petabyte scale; the continuity that comes from a research-computing center with a 20+ year history; and the Vanderbilt institutional resources that have stood alongside the federal grants in supporting the work. That role is significant, and the platform would not exist in the form it does today without it. What Vanderbilt is not, and does not claim to be, is the originator of LStore’s development. The authorship belongs upstream — with Dr. Beck’s research group, with the federal investment that funded that research, and with the REDDnet effort that bridged research into production. Getting that credit right is not a diminishment of Vanderbilt’s role; it is the precondition for representing the platform honestly.

The funding pattern across that whole arc is worth seeing as a single picture. From the original IBP and exnode research through REDDnet and into the current CMS Tier 2 production era, the work has been carried by sustained federal investment — the National Science Foundation for the upstream IBP research at UTK and for REDDnet at Vanderbilt, and the Department of Energy through the CMS-HI program for the petabyte-scale production use that has carried the platform since 2011 — paired at each stage with institutional resources from the universities that housed the work. That combination is what kept the project alive across handoffs between research groups, across the move from research network to production system, and across more than two decades of continuous development.

Dr. Alan Tackett is ACCRE’s Technical Director and the lead developer and chief architect of LStore. The platform’s architecture — the layered design, the placement model, the reliability machinery that the technical pages on this site describe — reflects his work and that of the team of developers he leads at Vanderbilt, together with collaborators at other institutions. When this site describes an architectural decision in LStore, it is describing a decision made by Alan Tackett and the ACCRE team, not by Unique Checksum.

Hunter Hagewood is ACCRE’s Interim Executive Director and Director of Operations, and shares in the architectural authorship of the platform. Lindsey Fox is ACCRE’s Director of Research Services, Facilitation and Outreach. Together with Alan Tackett, the ACCRE directors oversee the technical, operational, and research-services dimensions of LStore. The platform is an institutional achievement of Vanderbilt and ACCRE, sustained by a team over a long period, rather than the work of any single hand.

The research foundation beneath their work has its own authors, credited in the research lineage above: Micah Beck, Terry Moore, James S. Plank, Scott Atchley, and their collaborators at the University of Tennessee, whose logistical-networking research produced the Internet Backplane Protocol and the exnode that LStore is built on. Honest provenance credits both layers — the researchers who established the foundation, and the ACCRE engineers who built a production platform on it. Neither claim diminishes the other; the platform exists because of both.

The distinction is one the firm holds deliberately, because provenance done honestly requires it. The work Unique Checksum performs is real and substantial: it produces the technical documentation that gives evaluators a complete public reference; it builds the support infrastructure that institutional users depend on; it develops the materials that introduce LStore to organizations that have never encountered it; and it coordinates the relationships and integration that turn a research-engineering project into a platform an institution outside Vanderbilt can adopt, deploy, and rely on. That work is what the firm contributes. It is the work of carrying the platform forward, not the work of having created it.

Unique Checksum is the right firm to carry it because of its own lineage, which runs alongside LStore’s. The firm’s principal led Teradactyl for twenty-five years, building enterprise-scale data-integrity and backup technology for high-performance and research-computing environments — with ACCRE itself as a client since approximately 2005. The relationship that now produces this commercialization work rests on twenty years of shared operational history, not on a new commercial introduction. The firm understood ACCRE’s environment, its priorities, and its standards long before this engagement began. More on that heritage is on the About page.

The platform is open and freely available — an institution with the expertise to do so can obtain it and run it independently, as the engagement models page describes. This openness is consistent with the platform’s origins: LStore grew out of publicly-funded research and a research-computing center’s in-house engineering, and it remains accessible in that spirit. Unique Checksum’s commercial offering is the engineering, integration, and support that surround the open platform — not a gate on access to it.

Development continues at ACCRE, where LStore remains in active production and active refinement. The platform that an institution adopts today is not a frozen artifact; it is a living system with an engineering team behind it and a trajectory ahead of it. The provenance described on this page is the foundation of that trajectory, not a closed chapter — the accumulated depth on which the platform’s future development rests.

This is the healthiest possible position for an institution evaluating the platform: an open system with a deep and honest lineage, an active engineering home, named authors who continue to develop it, and an authorized commercialization partner who can deliver and support it without standing between the institution and the platform itself. The provenance is the assurance that LStore is real, proven, and here to stay.

For an institution weighing LStore, the provenance answers the question that matters most about any storage platform: can it be trusted with data that cannot be lost? The answer LStore offers is not a marketing claim but a lineage — logistical-networking research that established the architecture, more than two decades of ACCRE engineering that built and hardened it, years of production service carrying high-energy-physics and research data at petabyte scale, and named authors who continue to develop it. A platform with that history behind it is a different proposition from a product with a launch date.

Unique Checksum’s part in the lineage is to extend it — to carry a platform that has proven itself at Vanderbilt to the institutions elsewhere that need what it offers, with the same respect for the platform’s authorship and the same discipline about its integrity that the rest of this site reflects. The provenance is the foundation; the commercialization is the continuation. Both are presented honestly because honesty about provenance is, finally, part of being trustworthy with data.