RedCLARA built a continent-scale backbone through federation, not ownership

Summary

• RedCLARA is a non-profit association registered in Uruguay that federates national research and education networks. It is neither a shareholder-owned company nor a telecom operator that controls every campus connection.
• Eight major routing hubs and long-term international capacity form a 100G-class backbone, with 100, 200 and 300 Gbps circuits and a shared 20 Gbps segment in Central America reported.
• ALICE and ALICE2 built the organisation and its first network, and BELLA secured long-term capacity on a direct Portugal–Brazil route, reducing dependence on transit via North America.
• The test for BELLA II is to turn project funding into a base that members sustain; to strengthen weaker national networks and public evidence; and to prevent bottlenecks in supplier or membership structure.

BELLA changed the geography of the network

BELLA – Building the Europe Link to Latin America – combined a submarine element with terrestrial reinforcement. BELLA-S secured long-term optical spectrum on a direct Portugal–Brazil submarine cable, and BELLA-T strengthened the South American routes that carry capacity from landing points to regional nodes and national networks.

The physical system is EllaLink, the roughly 6,000 km cable linking Sines in Portugal with Fortaleza in Brazil. Research and education traffic went into production in 2021, when two 100 Gbps links were launched – one for GÉANT–RedCLARA and one for Copernicus. The 35,830 km and Nx100G reported in 2024 refer to the broader BELLA infrastructure, including terrestrial and international segments, not only the submarine section.

Distance and cost measure different scopes. CORDIS lists a total of approximately €12.44 million for BELLA-S1, with an EU contribution of €10 million; other totals include terrestrial work, additional funding and national contributions. They must not be merged without explanation.

Some materials show latency reductions of up to 50% on specific routes, which cannot be generalised to all institutions. The long-term achievement is the direct South Atlantic route and a right that can be expanded as optical technology and demand evolve.

40 optical channels do not mean submarine cable ownership

BELLA obtained long-term rights to 40 optical channels on EllaLink. Spectrum rights allow control over how capacity is lit and upgraded, but they do not make RedCLARA or GÉANT the sole owners of the commercial submarine cable, the landing stations or the related facilities.

There is a chain from fibre pairs to spectrum rights, optical channels and coherent transmission beyond 100G, and from landing stations to terrestrial networks. The submarine section is maintained by the cable operator, other organisations run landing and colocation, RedCLARA and GÉANT coordinate academic capacity, equipment vendors supply hardware, and the NRENs deliver to the institutions.

Spectrum rights can be more valuable than fixed circuits because, within optical and contractual limits, capacity can be increased through equipment upgrades. They also carry shared governance, repair, power, spectrum planning and interoperability responsibilities. Even long-term rights are not self-operating.

This boundary matters. Because RedCLARA genuinely coordinates and operates capacity, it can speak credibly about the regional backbone and transatlantic routes, but it should not be described as owning every line on the map. Its strength is not full physical integration; it is federal cooperation and contractual control of strategic capacity.

BELLA and the direct South Atlantic route

The clearest change BELLA brought to European–Latin American research connectivity is geographic. Previously, even traffic between the two shores of the South Atlantic could pass through North America. Sines–Fortaleza created a shorter option for many paths and left the research community a long-term asset.

The benefit is not that latency is halved on every path. Results vary by source, destination, national access, congestion and return path. “Up to 50%” should be limited to specific routes. A more robust outcome is diversity: European connectivity no longer depends only on North American gateways.

Large-scale data transfer, astronomy, Earth observation, distributed computing, academic telemedicine and interactive collaborative research benefit from a more direct and predictable path. RedCLARA and GÉANT can offer research interconnection that is easier to control than the general internet.

A direct submarine cable is still exposed to cuts, power failures, landing station, optical equipment and maintenance faults. Resilience requires alternative paths, switchover testing and enough capacity to use Miami and other gateways when EllaLink is down. “Direct” is an improvement, not a substitute for redundancy.

BELLA is also a long-term institutional relationship between Europe and Latin America. Because the capacity was secured through a science cooperation programme rather than a simple commercial transit purchase, the research community has some influence over use and upgrades. Physical operation, however, depends on commercial partners.

Spectrum rights, terrestrial reach and ownership boundaries

The phrase “RedCLARA network” hides several forms of control. The organisation owns routers, rents racks and power, purchases operational services, holds IRUs, uses optical spectrum on third-party cables and exchanges capacity. Each form differs in term, maintenance, upgrade rights and risk.

At sea, BELLA secured long-term spectrum on EllaLink. On land, BELLA-T and the members combined national fibre, wavelengths and services to reach PoPs and institutions. The 35,830 km in the annual report is the programme’s overall reach, not a distance owned entirely by RedCLARA.

This distinction prevents two errors. The strategic value of long-term spectrum rights should not be underestimated, but neither should it be overstated as RedCLARA single-handedly managing all cable laying, landing stations, submarine repairs, optical equipment and national contracts.

The real strength lies in the contracts. Term, maintenance responsibility, upgrades, operator acquisition or failure, landing station access, exit rights and alternative paths matter. Lines on a map only mean something once these conditions are understood.

A federal model can create broad reach without owning every physical asset. In exchange, supplier management and transparency become critical. It is essential to know which dependencies are replaceable and which have become irreversible single points of failure.

Continent-scale, but not a monolithic single operator

Calling RedCLARA simply “Latin America’s research network” makes it sound as though a single organisation directly controls everything from the international backbone to a university’s network socket. In fact it is federal. A researcher typically enters through the institution’s network, into a national research and education network (NREN), and through RedCLARA’s regional and intercontinental layer, connecting to separately operated national systems as part of a larger scientific environment.

This distinction shapes both value and limits. Brazil’s RNP, Chile’s REUNA, Colombia’s RENATA, Ecuador’s CEDIA, Mexico’s CUDI, Costa Rica’s RedCONARE and Uruguay’s RAU keep their own national governance, infrastructure, institutional relationships and operational responsibilities. RedCLARA does not replace them; it interconnects them, represents shared demand, coordinates capacity and services, and provides a path to GÉANT, Internet2, CANARIE and others.

A typical end-to-end path crosses the campus LAN, the institutional firewall, a national access circuit, the NREN backbone, RedCLARA, international peering, the counterpart NREN and the remote research facility. A regional 100 Gbps circuit does not remove campus egress congestion, slow national circuits, poorly configured storage or counterpart limits. Conversely, a single national NREN cannot negotiate all the intercontinental routes, trust federations and shared services that modern science requires.

It is therefore appropriate to describe RedCLARA as a shared coordination and infrastructure layer. “Shared” refers to governance, technology, long-term capacity rights and common services – not to ownership of every fibre, router, PoP and submarine segment on the map. This is precisely how multiple national systems and commercial infrastructure contracts are assembled into a regional public network.

Why Latin America needed a regional layer

The starting point was fragmentation. In the early 2000s, several countries had national academic networks, but no enduring continental organisation existed to aggregate demand, operate a regional backbone and negotiate on equal terms with European and North American research networks. International circuits were expensive, telecom markets differed sharply by country, and commercial paths were designed around operator economics rather than distributed science.

As a result, even a university with good national connectivity could end up paying for expensive, roundabout international transit to collaborate with another Latin American institution. Europe–Latin America traffic went via North America because that route was easier to procure in the commercial submarine cable and interconnection market. The problem was not only latency: it was weak bargaining power, a lack of path diversity and dependence on priorities outside the research community.

A centralised, single-operator model for the continent was not realistic either. NREN maturity, public funding, regulation, national fibre, university systems and international landing points differ by country, and each national organisation needed to keep its legal and operational autonomy. The federal model was therefore chosen: national networks keep their national role, and a regional body carries the common layer and the relationship with the wider world.

A single national NREN can hardly drive international research programmes or justify dedicated routes, but a regional organisation can pool scientific demand, bundle contributions, combine grants with national assets and face the European Commission, GÉANT, development banks, operators and submarine cable carriers as one counterpart. RedCLARA was created to turn that common demand into an organisation and a working network.

Toledo, Valle de Bravo and the legal entity

The institutional idea took shape in 2002. Latin American NREN representatives examined the possibility of a regional network at a CAESAR meeting in Toledo, Spain, and continued consultations in Rio de Janeiro, Buenos Aires, Santiago and Mexico. The questions were not only about topology: they covered eligibility, the relationship between a regional organisation and national networks, decision-making and how to turn externally funded projects into a permanent organisation.

The legal history has three dates. RedCLARA’s own history says the statutes were signed on 9 June 2003 in Valle de Bravo, Mexico. The Organization of American States’ civil society record gives 10 June as the founding date. RedCLARA also states that it was granted legal existence under Uruguayan law on 23 December that year. These should not be merged into one day as a contradiction; they should be treated as related but distinct events: the founding agreement, the administrative incorporation record and the later legal recognition.

The most accurate description is a non-profit international civil association based in Montevideo and recognised under Uruguayan law. It is sometimes described in English as an “international law organisation”, but there is no evidence that it is an intergovernmental body created by treaty. It is a membership organisation with regional purposes – not a supranational regulator and not a government agency.

The current statutes were approved in November 2019 and published in January 2020. The General Assembly is the highest body, with a Board of Directors, a fiscal oversight body, a technical committee of NREN experts and an executive team responsible for implementation. Legitimacy comes from membership and cooperation, not from shares or territorial authority.

Seven legal members, 11 connected countries and a broader mission

The current public membership list shows seven NRENs: RNP, RENATA, RedCONARE, REUNA, CEDIA, CUDI and RAU. The current advanced-connectivity page, by contrast, says that 11 countries are connected: Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Mexico, Paraguay and Uruguay. Older material mentions 13 countries. These are not the same metric.

Legal membership determines governance rights, obligations and participation in the association. Operational connectivity can exist through partners, transitional arrangements or projects even when the NREN concerned is not a formal voting member. Past country counts may reflect former members or earlier project scope. “Seven current legal members” and “11 connected countries” must therefore be kept distinct, and historical figures should be treated as point-in-time.

This gap creates strategic tension. RedCLARA supports a wider region than the membership body it formally governs. A country may benefit from connectivity and BELLA II without having equal standing in the General Assembly. Conversely, the seven members carry a heavier share of governance, technical work and ongoing costs. This is not proof of institutional failure, but it affects representation, cost sharing and the construction of a genuine regional platform.

The 2025–2027 Board has José Palacios of REUNA as President, Eduardo Grizendi of RNP as Vice-President, Juan Pablo Carvallo of CEDIA as Treasurer, Carlos Gamboa of RedCONARE as Secretary, Eduardo Grampín of RAU as Director, and Luis Eliécer Cadenas as Executive Director. While the national networks hold formal control, recent General Assembly minutes, voting ratios, fees and decision-making participation by connected non-member countries are not sufficiently public.

ALICE supplied the regional network’s initial capital

RedCLARA’s legal incorporation and first production network were inseparable from ALICE, the Europe–Latin America interconnection project. The European Commission and DANTE signed a contract in June 2003 under which the Commission contributed €10 million and the Latin American side €2.5 million of a €12.5 million total. With extensions, it ran until March 2008.

ALICE solved several start-up problems at once: funding, joint procurement, an immediate operational mission for the new organisation and connection to GÉANT. Rather than waiting for a mature membership and revenue model, it demonstrated network value while the organisation was still being built.

The first infrastructure was smaller than today’s, but institutionally significant. It created production connections between regional PoPs and the NRENs and established relationships with European and North American research networks. Those relationships later extended into identity, middleware, grid computing, open science and joint projects. A working network made the association more than a meeting body.

At the same time, ALICE left a hard problem: how to keep running infrastructure built with time-limited grants once the grant ends. Contracts, equipment, staff and user expectations do not disappear with the final report. Much of RedCLARA’s subsequent history is an attempt to acquire long-term rights – IRUs, dark fibre, wavelengths, spectrum, services and co-investment – to bridge funding cycles.

ALICE2 moved the centre of coordination to Latin America

ALICE2 began in December 2008 under an €18 million European Commission contract. The biggest institutional change was that RedCLARA became the coordinating body, moving from entity in a European-led programme to regional operator and project leader.

The programme explicitly aimed at reducing recurring costs through dark fibre, wavelengths and IRUs; expanding reach; strengthening NRENs; technical training; user communities; and sustainable funding. Sources contain both a 2012 end date and an administrative closure in January 2013. The safe reading is that operational activity ended first and administrative completion came later.

ALICE2 also broadened the definition of infrastructure. Grid computing connected regional resources to international collaboration, and identity and middleware improved the use of common services. Training recognised that optical circuits alone have little value without people, governance and applications in the national networks.

The programme moved RedCLARA from a young transmission network to a regional e-infrastructure organisation. Dependence on external funding remained, but it gained experience in coordinating large projects, managing multi-country technical work and turning transmission into a base for services.

From short-term circuits to long-term capacity rights

The economics of the regional network depend on how capacity is acquired. Short-term managed circuits deliver a clear service, but they are exposed to renegotiation and price volatility and leave little freedom to scale. Long-term rights to fibre, wavelengths and spectrum require upfront investment and governance, but allow planned expansion through equipment upgrades instead of buying bandwidth again each time.

Where it could, RedCLARA moved towards IRUs, dark fibre, wavelengths, optical channels and capacity exchange. It did not become the traditional owner of every cable, but it secured rights that remain useful beyond a generation of technology. A 2015 document linked this approach to lower variable bandwidth costs and sustainability.

Long-term rights also change risk. They reduce repeat purchases but deepen dependence on cables, landing stations, operators, maintenance and contracts. Spectrum has no value without repair, power and operations. Optical transmission equipment, colocation, terrestrial extensions and skilled staff are also required.

The asset base therefore does not reduce to a simple ownership map. NREN-owned fibre, leased segments, in-kind contributions, IRUs and optical rights on shared submarine cables are mixed together. The regional platform is assembled from legal and technical relationships as much as from physical assets.

Before BELLA, Europe usually meant going through North America

The absence of a high-capacity direct route between Europe and Latin America shaped both performance and dependency. Even between institutions on opposite shores of the South Atlantic, commercial traffic could pass through North America. That was rational for commercial market structures, but it added distance, concentrated transit and weakened scientific control over capacity.

North American connectivity was and remains essential. WHREN/LILA and similar efforts strengthened routes to the United States, including Tijuana, San Diego and Pacific Wave, and Miami became the gateway to Internet2, CANARIE and others. These provide reachability and redundancy.

The problem was not the existence of a route but over-dependence. A continental network needs alternatives. A direct European route lowers latency for specific pairs, spreads failures and keeps large scientific data from concentrating in a single corridor. It also gives physical expression to the long cooperation between RedCLARA and GÉANT.

Growing volumes of Earth observation imagery, genomic, telescope and climate model data and HPC workloads increased the need, and BELLA turned the direct-connection idea into a long-term physical and contractual system.

Current topology: eight main nodes and several forms of capacity

RedCLARA’s current technical descriptions list eight main routing nodes: Buenos Aires, Fortaleza, Porto Alegre, São Paulo, Santiago, Manta, Panama City and Miami. South American NRENs connect to these and are joined in rings of 100 Gbit/s or more. The shared Central American infrastructure is put at 20 Gbit/s. Fortaleza is the main direct point toward GÉANT and the Europe–Africa direction, and Miami is the main gateway toward Internet2, CANARIE and North America.

The advanced-connectivity page also lists links of 100, 200 and 300 Gbit/s, eleven connected countries and more than 1,300 institutions. The same site, however, still contains older wording with 10 Gbit/s. Newer topology and project material supports a 100G-class network, and the 10G wording should be treated as legacy content.

Advertised speeds do not mean that every path has the same usable capacity. Line rate, protected aggregate capacity, reserved capacity and the bandwidth actually offered to members differ. Where logical rings share commercial fibre or the same physical path, logical redundancy does not guarantee full physical separation.

Manta serves the Pacific side, Fortaleza the South Atlantic side, and Panama and Miami the Central America, Caribbean and North America sides. The structure is shaped by landing stations, national telecom markets, NREN relationships and existing fibre; it is not a uniform continental ring.

The methodology behind the figure of more than 1,300 institutions has not been published. It is unclear how many are actually in service, at what speed and which regional services they use. Maturity should be assessed through a combination of capacity, traffic, availability and service usage.

AS27750, peering and traffic classes

AS27750 is RedCLARA’s public autonomous system number. Through it, the organisation carries and exchanges routes for member networks, research and education peers and international services. Public databases show relationships with GÉANT, ESnet, Internet2, CANARIE, RNP, REUNA, RENATA, RedCONARE, CEDIA, TENET and others, but this is a point-in-time snapshot, not the complete contractual topology.

Research networks are not merely “a faster general internet”. Eligible research and education traffic can use dedicated or private paths to NRENs and scientific facilities and, where appropriate peering exists, avoid general commercial transit. This helps with cost, predictability and capacity reserved for research use.

At the same time, RedCLARA also provides commercial internet breakout, peering and global connectivity. Universities need both research resources and the public internet, and the two traffic types can have different policies, suppliers and cost models. A research network is not physically isolated from the public internet; it is distinguished by eligibility, routing, capacity and governance.

Selective peering is a policy of connecting directly wherever mutual benefit exists; it does not guarantee that every network will participate. Even in the same facility, ports, cross-connects, filters and operations are required.

Current RPKI validation, ROA coverage, IRR filtering, route-leak mitigation, BGP communities and per-peer availability are not fully public. More routing security evidence, to the extent confidentiality permits, would allow a more accurate assessment of the resilience of the regional infrastructure.

Layer-2 circuits, operations and the limits of regional responsibility

RedCLARA offers dedicated Layer-2 virtual circuits. Through VLANs, laboratories, data centres and observing instruments in different countries can be connected as if on the same Ethernet segment. This is valuable for high-volume data, distributed experiments and workloads that do not fit ordinary routed IP.

A single circuit crosses multiple administrative domains. The institution provides the local port, the NREN carries it to the regional network, RedCLARA coordinates the international segment and the counterpart handles the reverse flow. VLANs, MTU, redundancy, bandwidth, addressing, security and change windows must all be aligned.

Responsibility is distributed. A fault can occur at the campus, the NREN, RedCLARA, a transport provider or the counterpart institution. A regional NOC needs circuit identification, 24-hour contacts, demarcation of responsibility and escalation. If users only see “the network”, they cannot tell who can restore it.

RedCLARA describes engineering, NOC and security functions, but has not published the full SLA methodology, repair-time history or an independent incident register. The monthly average availability of 100% and 7.7 PB per year in older pages have no clear scope or method and cannot be treated as audited values.

In the federal model, RedCLARA coordinates the regional segment and works with NRENs, but cannot guarantee every element alone. Professional services need a way to make responsibility boundaries visible to users and to make the federation work as a joint support chain rather than a shifting of blame.

BELLA II: expansion through co-investment, not a finished map

RedCLARA and the European Commission signed BELLA II on 14 December 2022. It runs for 48 months with a total programme of €28 million, of which the Commission provides €13 million, while RedCLARA is expected to mobilise at least €15 million from governments, development agencies, companies, banks and others.

The targets are connectivity and strengthening for Peru, Costa Rica, El Salvador, Guatemala and Honduras, with possibilities indicated for Mexico, the Caribbean, Paraguay, Uruguay, Bolivia and Belize. These are goals or consultation topics; they do not mean production connectivity as of 3 August 2026.

The 2025 competitive dialogue sought a fibre system with a minimum lifetime of 15 years, 20 Gbit/s for Central America, 100 Gbit/s or more for Peru and future expandability. Existing fibre, new construction, capacity exchange, co-construction and combined options are all possible.

Flexibility can attract investment, but it also complicates accountability. It must be clear who builds, owns and maintains; how much capacity is for research; how countries participate; and who pays for operations after the grant ends. A 15-year right is only sustainable if power, equipment, repair and renewal are funded.

BELLA II should be evaluated on working routes, enforceable rights, accepted traffic and post-project operations. Memoranda and policy dialogue are progress, not service launch.

From a transport network to a regional research platform

Early RedCLARA showed that national research networks could exchange traffic over a regional backbone. Since then it has sought to build identity, file transfer, collaborative environments, training, HPC, open science, cybersecurity and thematic communities on top of it.

This is not a departure from networking; it is an attempt to add value. A fast path produces no research outcomes if users cannot log in, discover computing resources, move large data or find partners and support. The application and community layer turns bandwidth into usable research infrastructure.

BELLA II institutionalises this idea through innovation hubs, testbeds, open calls, training, strategic dialogue and the search for funding and partners. The 2024 agri-food Ideathon drew 125 people from 28 countries, and the early HPC testbed built a path from training and prototyping towards shared computing.

Testbeds are not production services. They typically have selected entities, limited capacity, specialist support and an experimental purpose. They can reveal faults before permanent investment, but presenting a successful demonstration as universal availability would mislead.

RedCLARA’s opportunity is to become a federation of services and experiments, not only transport; but identity, storage, computing, notebooks and sensitive data add security, privacy, support and lifecycle responsibilities. The underlying resources are operated by NRENs, SCALAC, universities, clouds and partners, while RedCLARA provides integration and the regional entry point.

BIO+IA and the limits of early-adopter evidence

BIO+IA Early Adopters is a clear example of this new platform role. It began in 2026 under BELLA II, selected five projects from six countries and provided training, specialist support and AI and bioinformatics environments. The topics were breast cancer biology, molecular screening, antimicrobial or mercury resistance, drug repurposing for liver disease and translational biomedicine.

At the July 2026 Demo Day, participating teams reported that they had received support in searching, integrating and interpreting biological information. This is evidence that the network, specialised tools and experts can be combined at regional level to support data-intensive biomedical research.

It does not, however, demonstrate clinical efficacy. Finding candidate molecules is not proof of therapeutic effect, and faster analysis does not complete the experiments, regulation or clinical adoption. The cohorts are limited, and the results are reports from entities and organisers, not an independent clinical trial.

This limitation does not weaken the result; it identifies the correct unit of outcome. RedCLARA supported a structured multinational experiment, lowered barriers to tools, data and expertise, and formed relationships among researchers.

The next test is continuity. A demonstration becomes infrastructure only when teams can reuse it, new cohorts can join, data governance is clear, computing and storage are funded, and successful methods migrate into national and institutional systems. Continued use, publications, independent evaluation and post-pilot funding all matter.

The larger lesson is to measure innovation programmes by the distance from participation to sustained capability. An Ideathon can count entities and a testbed can count experiments, but infrastructure must be counted by the maintained environment, continuing users, governed data, trained operators and projects that survive beyond the grant or event.

Identity as infrastructure: FIEL, eduGAIN and eduroam

A network path is of little use if you cannot prove your identity to the service at the other end. FIEL supports national identity federations; eduGAIN enables metadata exchange between federations and the use of international services with home-institution credentials; eduroam provides similar roaming for network access.

These are not a single regional password database. In a typical federated login, the service sends the user to their home institution, the institution authenticates them and returns a signed assertion or the required attributes, and the service applies its access policy. RedCLARA connects federations and international trust, but it does not usually hold passwords for all users.

Both trust and risk are distributed. Universities keep account responsibility, and researchers can use the same institutional ID for repositories, computing environments and collaborative tools. At the same time, assurance levels, attribute semantics and metadata freshness differ, and a compromise of an IdP or signing key can ripple across many services.

eduGAIN solves discovery and metadata exchange, but not every legal problem. Services decide which institutions and assurance levels to accept, and institutions manage keys, account lifecycles and incidents. Cross-border use requires consent, attribute minimisation, privacy and a legal basis.

With eduroam, the visited site provides the connection and forwards authentication requests to the home institution through the RADIUS hierarchy. Security depends on device configuration, certificate validation, the home institution’s authentication and the visited site’s controls.

Identity is continuity of infrastructure. Even if a supercomputer is reachable over IP, its intended researchers cannot use it if the trust path is broken. Coordinating connectivity and identity together is part of RedCLARA’s value.

MiLab, eNVIO and the collaborative research layer

MiLab brings together several open-source tools behind federated identity. It combines a Dataverse-based data environment, GitLab-based development, Mattermost communication and notebook computing around JupyterHub, Python and R, supporting communication, code, analysis, data curation and publication.

Distributed teams normally stitch together separate services with different accounts, data policies and support models. A regional workspace reduces fragmentation and gives researchers in multiple countries and institutions a common environment. RedCLARA documents collaborative research using geographically distributed data, among other cases.

MiLab should not be called a universal regional cloud. It does not automatically provide large-scale HPC, unlimited storage, approval for every kind of sensitive data or legally permanent preservation. Capacity, service levels and hosting can change, so fit with project requirements is necessary.

eNVIO addresses the problem of moving large files on academic infrastructure. Email and consumer-oriented sharing are unsuitable for much scientific data. Its value, however, depends on size limits, retention, encryption, support and NREN integration, and the full metrics are not public.

RedCLARA also operates a regional training platform and an LMS with CEDIA. Training spreads expertise but does not replace permanent posts, real operational experience or institutional responsibility.

This collaborative layer is a microcosm of the federal model. RedCLARA integrates tools and the entry point; NRENs connect users; institutions provide identity and context; researchers do the work. It works when responsibilities are clear, but regional services cannot compensate for every gap in local staffing, data governance or application support.

Ventanilla Abierta and navigating a fragmented ecosystem

Regional research infrastructure can exist and still be hard to find. Researchers may know that HPC, storage, identity and specialised networks exist somewhere in Latin America, but not know who runs them, which projects are eligible or where to start an application. Ventanilla Abierta was created as a one-stop guidance layer for this problem.

The service does not own all the resources to which it refers. It translates research requirements into routes: connectivity through NRENs, computing through SCALAC and others, data and storage, identity federations, international partner resources, and projects and funding. Its value lies in coordination and discoverability.

This is a form of navigation infrastructure. Physical capacity may sit unused not because there is no demand, but because the institutional windows are fragmented. When multiple NRENs, programme rules and technical teams have to be understood separately, regional resources are underused. A trusted entry point lowers transaction costs.

Funds and Partners addresses the related problem of opportunities, complementary institutions and money. By connecting technical requirements with partners and funders, it can turn potential ideas into projects that can use the infrastructure.

Its mandate has limits, however. Advice is not a reservation, a call is not an award, and an introduction is not a consortium contract. Success should be measured not by the number of enquiries but by resources actually connected, collaborations formed and projects that continue.

LA Referencia, SCALAC and the shared knowledge and computing chain

RedCLARA’s relationship with open science and HPC shows how transport becomes an ecosystem. LA Referencia aggregates metadata and research outputs from national repositories, and SCALAC links advanced computing organisations. Neither is a subsidiary of RedCLARA; both are independent collaborations, but the regional network and its coordination support cross-border operation.

The open science chain starts with institutional and national repositories. Researchers deposit papers and data under local policies, national systems curate metadata and access, LA Referencia provides regional discovery and interoperability, and relationships such as those with OpenAIRE extend discovery internationally. RedCLARA supports connectivity and institutional coordination but owns none of the repositories or outputs.

Open science requires more than bandwidth. Metadata quality, persistent identifiers, licences, preservation, repository governance and researcher incentives determine reusability. A fast network can carry files, but it cannot make undocumented data reproducible.

SCALAC faces the same problem in computing. HPC resources are dispersed across universities and national centres. Regional collaboration makes facilities and experts visible and can produce testbeds and training that would be difficult for any single country alone.

A regional HPC federation is not one supercomputer. Schedulers, software, access rules, queues, storage and charging differ, and data placement, identity and allocation governance are required; sensitive data also faces cross-border constraints. The network removes one bottleneck and makes the remaining problems visible.

The strategic value is choice. Countries can participate without owning every specialised system, and facilities with spare capacity gain collaborators. The risk is that use remains dependent on projects and personal relationships instead of becoming a published, predictable service with clear eligibility, capacity and costs.

Health, Earth observation and the communities that give the network purpose

RUTE-ALC connects a community of university telemedicine and digital health. Its scope is understood to include specialist consultation, medical education, image exchange, standardisation, emergency cooperation and digital health governance.

A research network can provide trusted paths, identity and collaborative tools, but it does not become a healthcare provider. Clinical privacy, medical device regulation, professional liability and patient consent remain with health institutions and public authorities. A fast circuit is one condition for telemedicine, not the complete clinical service.

The regional activity around the Copernicus Academy builds another bridge. European Earth observation data can be used in agriculture, disaster response, climate adaptation, oceans, water, forests, urban planning and energy. RedCLARA supports data movement and training, but the constraint is often analytical capacity rather than access to raw imagery.

TICAL brings together IT leaders and technical communities in higher education. The 2024 meeting in Rio de Janeiro drew more than 200 entities, according to the annual report. A conference is not physical infrastructure, but it aligns procurement, security, cloud, identity and service priorities and supports later interoperability.

These communities show why a research and education network is not simply a discount telecom operator. Continuing relationships speed incident response and project formation. But participation is not evidence of implementation; training does not automatically create national capacity; and connectivity does not prove health or environmental outcomes. RedCLARA sets the conditions for action and coordinates.

EduLACSeg and a regional cybersecurity trust network

Because academic institutions face similar attacks across borders, cybersecurity cooperation is a natural extension of a federal network. eduLACSeg provides a trust framework in which RedCLARA and the NRENs share indicators, context and response experience. The value is early notification to many countries of an attack first observed in one.

In July 2026, threat information originating from CERN was shared through RedCLARA to RNP, REUNA, RENATA, RedCONARE, RAU, CUDI and CEDIA. This is concrete evidence that the regional layer can carry security information. RedCLARA has described eduLACSeg as the largest academic cybersecurity ecosystem in Latin America, but with no independent comparative definition it should be treated as self-assessment.

Threat sharing is not automatically safe or accurate. Indicators carry errors, age and context dependence; attribution is uncertain; and rules on personal data and confidentiality differ by country. Some NRENs have mature security teams; others depend on very small ones. Sharing proves distribution, not uniform response.

The 2024 report records that a training with LAC4 drew 85 people from around 60 organisations in 18 countries, with a 30% share of women, and that a Security Baseline Bootcamp was held at TICAL2024 with RNP and GÉANT. These are capacity-building indicators, not evidence that identical controls have been deployed.

RedCLARA’s own routing, identity metadata, collaborative platforms and testbeds are also assets to defend. Its full current policies on RPKI, ROAs, IRR filtering, incident disclosure and response times are not public. Long-term trust requires rules for classification, redistribution, data protection, attribution and post-incident learning, together with measurable technical controls.

SUBMERSE and the network as a scientific instrument

SUBMERSE explored the possibility of using the communication fibre itself as a sensor. Distributed acoustic sensing analyses optical pulses and backscatter to detect the tiny perturbations that earthquakes, ocean movements, traffic and other events produce along the fibre.

Existing submarine and terrestrial fibre passes through places where deploying a dedicated sensor network would be expensive. A research network can connect scientists, cable owners, landing stations, data and computing resources, and can turn existing infrastructure into an observation base.

Fibre does not, however, automatically become an observatory. Measurement equipment, access rights, calibration, time synchronisation, storage, processing and scientific validation are needed, and production traffic must not be harmed. The data may also reveal sensitive activity, so governance is required.

RedCLARA is one entity in a wider consortium; it is neither the owner of all the fibre nor the sole inventor of the technique. The main funded period ended with the final dissemination activities in 2026, so it is more accurate to describe this as a direction of demonstration and cooperation rather than a permanent service.

In the future, if access, standards, funding and users can be secured, fibre sensing could give RedCLARA a distinctive scientific role in geophysics, oceans and infrastructure resilience. The network would not only carry data; it would generate data.

An ecosystem of partners, not a collection of subsidiaries

RedCLARA’s relationships include national NRENs, the European Commission, GÉANT, Internet2, CANARIE, development banks, EllaLink, telecom operators, SCALAC, LA Referencia, RUTE-ALC, LNet/LACChain, universities and thematic communities. These legal and operational relationships are not all alike.

NRENs are members, operational partners, or both. GÉANT and Internet2 are peers; the European Commission is a funder and programme counterpart; EllaLink is the commercial infrastructure on which capacity rights sit. SCALAC, LA Referencia and RUTE-ALC have their own governance. Universities can be users, hosts, beneficiaries and providers of expertise.

Calling all of this RedCLARA would exaggerate central control and blur responsibility. The value lies in coordinating assets and organisations that it does not own. Membership, funding, capacity, interoperability, implementation, collaboration and project participation must be distinguished.

An ecosystem avoids dependence on a single owner, but it can also diffuse responsibility. When something fails, it matters that people can tell whether the institution, the NREN, RedCLARA, an operator, a platform or an international partner is accountable. A federation with visible boundaries is stronger.

Finances based on membership, projects and capacity rights

RedCLARA has no confirmed shareholders, no market valuation and no distributable profits. The statutes provide for membership fees and other resources; consulting services, EU and development bank projects, international cooperation, partner contributions, co-investment, capacity exchange and in-kind provision also count as resources.

The prominent figures are programme budgets, not organisational revenue. ALICE was €12.5 million, ALICE2 €18 million, BELLA consisted of several distinct actions, and BELLA II is a €28 million programme with €13 million from the EU and more than €15 million to be mobilised. These cannot be summed into RedCLARA’s turnover.

Part of the value is non-cash: optical spectrum rights, IRUs, national fibre, colocation, equipment and staff. It should be assessed by term, maintenance, upgrade rights and contractual constraints; nominal project amounts alone are not enough.

The 2024 annual report gives detail on activities and staff, but current full audited financial statements, reserves, membership-fee concentration, supplier risk and exact current headcount are not published. This gap constrains any assessment of sustainability.

The central risk is a mismatch of time horizons. Projects run for a few years; the backbone, identity and platforms operate indefinitely. Long-term rights retain value, but power, equipment, support and staff require continuing income. The key is to combine membership fees, cost recovery, national contributions and partners, and not to confuse grants with permanent revenue.

The last mile is a national and institutional responsibility, and it is uneven

However advanced the continental backbone, benefits can be uneven. Countries differ in national fibre, NREN funding, government support, telecom markets, landing stations, university capability and the supply of engineers. Within one country, a central university may be fast while a remote institution has fragile connectivity.

RedCLARA can aggregate demand, negotiate capacity, train and demonstrate the value of NRENs, but it cannot substitute for all national investment or regulatory problems, and it does not operate each access circuit. The last mile remains with institutions, NRENs and domestic providers.

So “a country is connected” does not mean every university receives the same quality. The figure of more than 1,300 institutions does not show methodology, bandwidth, availability or real usage. Mature measurement should separate registration, operation, speed and use of regional services.

BELLA II extends the scope and, at the same time, creates new obligations in places with weaker operational capacity. Sustainable inclusion requires national engineering, budgets, support, security and identity integration. An international high-speed link without a strong NREN can become an isolated high-capacity facility.

Documentation debt obscures operational maturity

RedCLARA’s public record is rich, but it contains seven members, eleven connected countries, an older count of thirteen, 100G-class descriptions alongside 10G wording, and figures for institutions, traffic and availability without a published methodology.

The most likely explanation is documentation debt: pages from different generations remain. That does not contradict the current network, but it forces readers to check sources, dates and definitions. In infrastructure, old documents can be an operational risk, not only an editorial one.

Topology should state dates, contracted capacity, usable capacity, protection scheme, operator, operational status and type of right. Traffic should state period, direction and aggregation method; availability should state scope and formula; and membership should distinguish governance, connectivity and project participation.

Consistent publication helps members, funders and users decide, and reduces the risk that memoranda and planned routes are mistaken for working circuits. Transparency is not the disclosure of secrets; it is the definition of claims.

RedCLARA’s achievement is federation at the scale of infrastructure

The central achievement is not a cable, a router or a project: it is a durable regional institution in which autonomous national networks aggregate demand, operate interconnection, acquire long-term rights, join international projects and build common services while retaining national governance.

ALICE turned the idea into a production network; ALICE2 moved coordination into the region and pursued sustainability. BELLA changed the geography with the direct South Atlantic route and spectrum; BELLA II seeks to extend the scope into identity, open science, computing, security and experimentation.

Federation also sets limits. RedCLARA does not own all assets, does not control each NREN, does not guarantee every last mile and does not have unlimited money. Its power is coordination: it is strong when the incentives of members, funders, peers and suppliers align, and weaker when responsibility is blurred or national capacity is thin.

Future value will depend less on the next speed record than on the quality of the institutions surrounding the capacity: representation, long-term contracts, financial transparency, measurable security, service lifecycles and the path from pilot to production. The backbone is the foundation; the federal institution is the strategic asset.

Indicators that BELLA II has become permanent infrastructure

The first indicator is operational performance. Memoranda, dialogue and procurement requirements must become named suppliers, accepted routes, production traffic, fault procedures and published dates. Capacity should distinguish initial minimums, line rate, protected aggregates and usable bandwidth. The 15-year target only means something if operations, maintenance and renewal rights also persist.

The second is membership structure. It must be tracked how the gap between seven NREN members and eleven connected countries changes, whether target countries become members, permanent partners or institutional entities, and how that is reflected in voting, fees, service rights and technical obligations. Expanding the network without widening the governance base increases burden and concentration.

The third is operational evidence: dated topology and traffic, SLA methodology, incident disclosure, routing security, RPKI, and measured diversity between EllaLink, Miami and alternative paths. A major outage or route leak will test the real resilience behind the logical rings.

The fourth is service conversion. BIO+IA, HPC, MiLab, identity, eduLACSeg and open science should be assessed not by entity numbers but by repeat users, production allocations, support, data governance and post-grant funding. A pilot becomes infrastructure when users can plan around it.

The last is finance. Public release of current audited accounts, reserves, membership-fee and supplier concentration, and the ratio of unrestricted to restricted funds would increase confidence. What will be decisive is whether operations can continue through European funding cycles without turning the weaker NRENs into temporary beneficiaries.

Evidence-based scenarios

In a successful expansion scenario, Peruvian and Central American NRENs enter long-term, expandable routes and the regional base widens. The direct European link, the North American gateway and Pacific-side access support large instruments and shared computing.

In a services federation scenario, identity, security, data, HPC and testbeds become everyday services and connectivity fades into the background. Value increases, but so do software, privacy and support responsibilities.

In a concentrated governance scenario, projects and connectivity expand while legal membership stays at seven. Operations may look regional, but representation, costs and succession rest on a small core.

In a project-cycle scenario, innovations start with grants and cannot be sustained after they end. The backbone survives, but higher-level services fluctuate with funding. Without a permanent cost model, this is the most realistic downside.

Co-investment with telecom operators can improve sustainability through capacity exchange, co-construction and long-term rights, but it deepens dependence on a small number of suppliers, so maintenance, exit and renewal must be strictly managed.

In a network-as-sensor scenario, SUBMERSE techniques extend onto regional fibre and become environmental and geophysical services. This is still a future picture, but it could create a distinctive scientific role that commercial capacity alone would not provide.

Who controls, and who carries the risk?

Control is distributed across several layers. Member NRENs govern the association through the General Assembly and the Board; the executive and technical groups carry out regional implementation. National networks control domestic facilities and institutional access; GÉANT and others control their own networks; funders control programme scope; and operators, cable, colocation and equipment companies control the physical dependencies. Universities and the scientific community create demand but do not necessarily hold regional governance rights.

Avoiding a single owner separates decisions from consequences. An NREN may carry the last-mile cost of a project the Board approved; a funder may set four-year targets while members inherit decades of maintenance; an operator may hold the only path into a country; and users may not know whom to call for repair. These boundaries are the main control surfaces.

Incentives should make sustainable contributions visible – membership fees, fibre, seconded staff, service operations, data management and security response – not only project participation. Countries receiving support need a path to domestic responsibility and continuing obligations; strong members need assurance that solidarity does not become an unlimited transfer of costs.

Supplier contracts should protect path diversity, renewal rights, audit, exit and continuity through changes of ownership or strategy. Long-term rights reduce price risk, but if landing station, optical equipment or maintenance relationships fail, the capacity can become unusable. What matters is not only the bandwidth purchased, but which dependencies become irreversible.

Decisions that cannot be postponed until the next grant

First, RedCLARA must decide whether it remains primarily a federation of connectivity or becomes the permanent operator of a common digital platform. MiLab, identity, security and AI testbeds create value, but they also add data, software, staff and liabilities. A permanent service needs an owner, lifecycle funding, incident response and termination rules.

Second, the gap between legal membership and operational scope should be closed. Connected countries need status, voice and a path to membership or another accountable framework. Informality is convenient during expansion, but it blurs who governs, who pays and who can leave.

Third, financial transparency must be used as a governance tool. Audited accounts, reserves, fees, restricted funds and supplier concentration determine whether the organisation can survive funding interruptions and cable failures.

Fourth, a continuity architecture should be published covering path diversity, incident and escalation procedures, configuration and registry backups, identity and platform recovery, cable alternatives and succession for project services. Essential services must survive the failure of a single grant, supplier, path or national institution.

The irreversible risk is building essential services and long-term dependencies faster than continuing authority, finance and operational evidence. The irreversible opportunity is turning BELLA’s rights, NREN governance and international relationships into a regional infrastructure commons that no single country could build. Decentralisation must be preserved while responsibility is made explicit.

The long-term significance of a shared regional backbone

RedCLARA demonstrates a form of infrastructure sovereignty through cooperation rather than isolation. Latin American science gains agency not because one organisation owns all the fibre, but because national networks jointly manage a regional layer, hold strategic long-term capacity and participate in international research relationships on more equal terms. Dependence on commercial infrastructure remains, but mission, path and upgrade decisions are not left to the general internet market alone.

This model is useful beyond research. It shows how institutions with fewer or uneven resources can pool demand without losing national autonomy, and how grants can be converted into rights and organisations that outlast the project. It also shows the limit: coordination alone cannot indefinitely compensate for weak national networks, opaque finances and unclear accountability.

The decisive choice is whether the next stage is seen as extending the map or maturing the institution. The number of countries and speeds matter, but the permanent assets are the systems of governance, operations, identity, data, security and scientific cooperation. If those become transparent, financially strong and genuinely regional, RedCLARA will remain one of Latin America’s most important shared digital infrastructures. If not, the backbone will still carry traffic, but the higher ambitions will keep depending on the next round of external funding.