Reintech's two routed /24s now expose two logical exits—not two proven paths out of Villavicencio

Summary

• REINTECH SAS presents a credible, compact operating footprint around Villavicencio rather than a nationwide backbone. Its own pages advertise fibre-to-the-home in the urban area, high-power radio for rural and farm locations, business connectivity and television. Public registration records connect the Colombian company to AS274248 and an IPv6 /32, while current route observations show two IPv4 /24s, the IPv6 aggregate and two IPv6 /48 more-specifics. These facts establish activity and address authority; they do not locate towers, fibre routes, aggregation nodes, interconnection rooms or protected power systems.
• The current routing picture is more complex than the original single-neighbour thesis. A dated global view now observes AS262191 and AS273103 adjacent to AS274248, and full-table path samples show both immediately before Reintech on most visible routes. Yet one IPv4 /24 appears only behind AS262191 in that sample, another public view still reduces the network to one visible adjacency, and no public material proves that the two logical neighbours use independent ducts, poles, radio hops, buildings, cities, power feeds or commercial arrangements. Two BGP exits are useful evidence of choice, not proof of two physical ways out of Villavicencio.
• Customers should separate allocated addresses, installed equipment, saleable access speed, currently usable throughput and survivable capacity. Reintech advertises residential plans up to 900 Mbps and a 1 Gbps dedicated business offer, but it publishes no upstream circuit sizes, oversubscription ratios, tower-channel inventory, OLT or splitter counts, backup runtime, spare-radio stock, failover tests or measured recovery intervals. The public evidence supports a Medium network grade: an active local operator with identifiable resources and two observed neighbours, alongside unresolved questions about route authorization for one /24, operational IPv6 at customer edges, physical path separation, power continuity and the authority chain needed to restore service.

At humid dawn, the first route is still inside the building

At a humid Villavicencio dawn, the most revealing place to begin is not a global route collector. It is the imagined edge of a real subscriber installation: an optical terminal on a wall, or an outdoor radio aligned toward a hill or rooftop. This is an analytical scene, not a claimed visit to a Reintech site. Its purpose is to expose the distance between an internet route and the physical path that makes that route usable.

Before a packet can reach either of the autonomous systems visible beside AS274248, it may have to cross household wiring, an optical drop or wireless hop, a neighbourhood distribution point, an aggregation switch, a backhaul circuit, a powered router and an interconnection that Reintech does not publicly locate.

Reintech's public service site gives that local edge a recognisable shape. It divides prospective customers between urban and rural service, describes Villavicencio's centre as a fibre-to-the-home area, and presents radio connectivity for surrounding municipalities, rural settlements and farms. Those statements are useful because they identify two access media with different failure behaviour. A cut or badly bent fibre drop is not repaired like a radio whose alignment, interference environment or power injector has failed. A passive optical branch can affect several premises from one damaged point; a point-to-multipoint radio sector can expose many users to one antenna, channel or tower-power problem. Neither the home page nor the ordering path discloses the actual topology behind those categories.

The distinction becomes clearer on Reintech's wireless-network service page. It advertises LAN and WAN work, mesh arrangements, point-to-multipoint links from a central node, high-capacity point-to-point links and spectrum analysis. That is evidence of technical capability and of the kinds of systems the business is prepared to sell or install. It is not an inventory of Reintech's own access network. The page does not identify which hills, towers or roofs carry customer traffic, which frequencies are used, how many sectors are active, whether any backhaul is licensed, how much spectrum is clear during peak hours, or whether a radio site has an independent fibre return.

This is why a route trace must start inside the building. The visible internet path is only the last part of the service chain. For an urban fibre customer, the early path likely includes an optical network terminal and some form of passive distribution before reaching powered access and aggregation equipment. For a rural radio customer, it likely includes customer-premises radio equipment, line of sight to an access point, and a backhaul toward the operator's core. Those are normal engineering inferences from the services described, not confirmation of Reintech's exact design.

The same caution applies to a business circuit: a labelled dedicated service can still share a building entrance, pole line, metro duct, router, upstream room or power source with other products.

The operating surface therefore has at least four layers. There is the subscriber edge, where power, cabling and equipment ownership may sit partly with the customer. There is the access layer, split publicly between fibre and radio. There is an aggregation and routing layer associated with AS274248. Finally there is the external connectivity represented by neighbouring networks in BGP. A failure at the first two layers may leave AS274248 perfectly visible to the rest of the internet while a village, farm, office or apartment remains offline.

Conversely, a local radio can remain aligned and powered while an external route or interconnection removes global reachability.

That layered reading matters for Reintech's audience. A family buying broadband usually experiences the chain as one service and one bill. A farm may depend on the link for messaging, payments, cameras or remote administration. A small company may place cloud access, customer communication and electronic transactions on the same circuit. A television subscriber may not know whether the video service shares any access or aggregation component with internet traffic. Public routing evidence can establish that Reintech participates in the global internet.

It cannot, by itself, tell any of those users which physical component is their single point of failure.

REINTECH's legal and operating boundary is young and narrow

The company boundary is clearer than the site map, although it still requires care. Reintech's about page describes an owned fibre network, more than 50,000 active users, complete fibre service and 99.9 per cent real availability. These are material claims because they suggest scale and an ambition to control access infrastructure rather than merely resell someone else's retail connection. They are also self-published claims. The page does not provide an audit period, a denominator for active users, a definition of availability, a service-area breakdown or an independent measurement record. “Owned fibre network” may describe some plant while leaving ducts, poles, backhaul, interconnection space and external transit under other parties' control.

The company identifies itself in its site footer as REINTECH S.A.S. with Colombian tax identifier 901551794-1. A ColombiaPymes company page associates the same business with Villavicencio and says it began activities on 22 December 2021, with activities including wireless telecommunications, television programming and transmission, and other telecommunications. That page is a commercial republication rather than an authoritative live company certificate, and it returned an access restriction during the final review. It is best used as dated secondary context, not as conclusive proof of current ownership, management or scope.

A separate Portafolio company listing repeats the tax identifier and places the company at Carrera 2 6 19, Vereda Barcelona, Villavicencio, Meta. The ColombiaPymes page gives Carrera 5 Este 15-49. These differing addresses could reflect a registered-office update, formatting, separate administrative and operating locations, or stale republication. Public evidence reviewed here does not decide among those possibilities. Neither address should be promoted into a network node, tower, headend or interconnection facility without site-level corroboration.

Network registration creates a second, more technical boundary. The LACNIC record for AS274248 lists the autonomous system as an active direct allocation to REINTECH SAS, registered on 11 November 2025 and changed the following day. It places the registrant in Villavicencio and provides an operational contact. The matching LACNIC IPv6 record assigns 2803:77d0::/32 directly to REINTECH SAS with the same registration date. These are strong evidence that the company has its own public routing identity and IPv6 resource. They do not show when the first customer used those resources, where the routers sit, or which corporate party owns each fibre, radio, rack and power system beneath them.

The dates suggest a young sequence. The secondary company record points to commercial activity from late 2021, while the autonomous system and direct IPv6 allocation arrived almost four years later. A business can provide connectivity before receiving its own autonomous system by using another carrier's addresses or routing arrangement. It can also obtain resources before fully moving customers onto them. Public routing history, examined below, shows announcements appearing after the November 2025 registrations.

That alignment supports the interpretation of an operating network maturing into its own routing identity, but it does not disclose the migration process or the degree of continued dependence on partner resources.

One should also resist treating the legal entity, the access brand and every technical asset as the same thing. REINTECH SAS is the named company. Reintech is the public-facing service identity. AS274248 is a routing identity controlled under a resource registration. Television is offered through a named integration partner. One visible IPv4 prefix sits inside a larger block registered to another company. External autonomous systems carry traffic toward or away from Reintech. These boundaries are normal for a regional provider, but they determine who can act during a failure.

Reintech may be able to replace a customer radio immediately while needing a partner to repair a backhaul, change a route, restore a cross-connect or correct an address authorization.

The evidence supports calling Reintech a real regional internet operator, not merely a marketing page. It has customer-facing products, a consistent tax identity, a public autonomous system, registered IPv6 space and live announcements. The narrowness lies in what can be demonstrated: Villavicencio is the clear centre of gravity, while the named extent beyond the city remains general; address records identify offices but not infrastructure; and externally observed routes disclose logical relationships but not the underlying contracts or physical assets. That is a meaningful operating boundary, but not a complete resilience boundary.

Five announcements are not five independent networks

At the observation point used for this assessment, the RIPEstat announced-prefix view reports five routes originated by AS274248: 38.191.213.0/24, 38.211.254.0/24, 2803:77d0::/32, 2803:77d0::/48 and 2803:77d0:1::/48. Counting them is straightforward. Interpreting them as capacity or diversity is not. The two IPv4 /24s together contain 512 addresses, including addresses that protocol rules reserve within each subnet; that number says nothing about how many are assigned, reachable, shared through address translation, idle or protected by separate paths.

The IPv6 count is even easier to misuse. A /32 allocation contains 65,536 possible /48 customer or site networks. That mathematical span does not mean Reintech has 65,536 customers, 65,536 access nodes or anything close to that physical scale. Address space is a numbering resource, not a statement of installed ports, fibre kilometres, radio sectors or throughput. The two /48 more-specifics are announcements drawn from the /32, not additional independent blocks.

Announcing the aggregate and more-specifics can support routing policy, traffic engineering or operational segmentation, but the public view does not establish the purpose in this case.

The two IPv4 resources have different registration contexts. An RDAP lookup for 38.211.254.0 resolves to the exact 38.211.254.0/24 network, named REINTECH-CGNT-NET-1 and registered to REINTECH SAS in Villavicencio. An RDAP lookup for 38.191.213.0 instead resolves upward to 38.191.192.0/19, named TV-MAS-SAS-CGNT-NET-1 and registered to TV&MAS S.A.S. Reintech is visibly originating the contained 38.191.213.0/24, but the public registration does not show the commercial or operational instrument under which it does so. It could be delegated, leased, routed for service integration or covered by another arrangement. The exact boundary remains unresolved.

The RIPEstat routing-history view gives a time-bounded sequence rather than a full operating diary. At its coarse observation resolution, the 38.191.213.0/24 and IPv6 aggregate become visible around late November 2025. The two IPv6 /48 more-specifics appear around January 2026, and 38.211.254.0/24 follows near the end of that month. Because the history is sampled and depends on collector visibility, these dates should be read as observed windows, not exact commissioning timestamps. Still, they fit the registration chronology and show a footprint assembled in stages.

One recent feature of the prefix history also needs restraint. The announced-prefix data shows a gap for 38.211.254.0/24 between 7 and 9 July 2026 under RIPEstat's threshold, which requires visibility at a minimum number of full-feed peers. The gap may reflect withdrawal, reduced propagation, collector visibility, maintenance or another routing event. It is not enough to declare a customer outage. A prefix can fall below a global observation threshold while some paths remain usable, and it can be visible globally while local access is broken.

Without Reintech incident records, active measurements from affected customers or corroborating route views, the gap is a question for the operator rather than an outage finding.

The prefix set therefore proves several useful things. AS274248 is not empty. Reintech originates both address families. It has added resources since its initial appearance. At least one exact IPv4 block and the IPv6 /32 are registered directly to the company, while another originated IPv4 /24 sits inside a partner-registered allocation. None of this establishes that the five routes traverse separate routers, circuits or sites. All five could converge on a common device or common outbound path; different routes could also receive different policies. The route count alone cannot tell.

For customers, the more relevant question is where services land within these announcements and how portable they are. A fixed public IPv4 address on one /24 may be harder to move during a route problem than a customer connection hidden behind shared addressing. An IPv6 customer prefix might be renumbered or rerouted differently from IPv4. A business using allowlists may have a longer recovery if its address changes. A television or managed-network service may use private addressing internally. Reintech publishes no public mapping from products to prefixes, sites or recovery options.

The disciplined conclusion is that there are five visible announcements and two independently registered address families, not five independent networks. Address evidence is strong for existence and chronology. It is weak for installed capacity and silent on physical diversity. Any claim of redundancy must come from the path, equipment and recovery design beneath those announcements, which is where the two observed neighbours become important.

Two visible neighbours change the question, not the answer

The live routing evidence no longer fits a simple statement that Reintech has only one visible external neighbour. The RIPEstat routing-status result, observed on 17 July 2026, reports two seen neighbours and very broad route-collector visibility: all 327 monitored IPv4 peers in that result saw the origin, while 321 of 322 monitored IPv6 peers did. Those are signs of wide propagation at that moment. They are not availability percentages, and they do not measure whether a Villavicencio subscriber could actually reach the internet.

The corresponding ASN-neighbours result, with neighbour data dated 16 July, identifies AS262191 and AS273103 on the left side of observed paths. “Left” describes where those networks appear in collected AS paths relative to Reintech; it does not label either relationship as paid transit, settlement-free peering, backup, customer, reseller or physical carrier. Contract terms and traffic direction cannot be recovered reliably from that field.

A deeper BGP-state sample, timestamped 17 July at 01:59:37 UTC, shows why the distinction matters. Across the collected paths reviewed, AS262191 is the only immediate predecessor visible for 38.191.213.0/24. Both AS262191 and AS273103 appear immediately before Reintech for 38.211.254.0/24 and for the IPv6 aggregate and /48 more-specifics. Path counts vary by route and collector, so they are not traffic shares. The pattern nevertheless suggests that the second logical exit is not uniformly visible across every announcement.

Another public view remains narrower. The CIDR Report for AS274248 shows one adjacency, AS262191, and depicts the two IPv4 routes behind it. The difference does not make either public service fraudulent. They use different collectors, refresh schedules and classification methods. It does mean that a customer should attach a timestamp and observation method to every topology claim. “Reintech has two observed neighbours in current RIPEstat data” is defensible. “Every Reintech service has two active upstream paths” is not.

The CAIDA AS Rank page for AS274248 adds another analytical perspective. Its graph-derived measures place the network near the edge, with two providers, no observed peers or customers, and a small customer cone containing two prefixes and 512 IPv4 addresses. These are inferred internet-topology relationships, not declarations by Reintech or counterparties. They reinforce the picture of a compact access network dependent on larger networks to reach the wider internet, while leaving commercial status and physical separation unresolved.

Registration records identify the neighbours but do not locate the handoffs. The LACNIC record for AS262191 associates that system with Liberty Networks de Colombia S.A.S. and a Bogotá registration address. The LACNIC record for AS273103 associates the second with TV&MÁS S.A.S. and a Popayán registration address. A corporate contact address is not an interconnection point. It cannot prove that Reintech traffic physically travels through Bogotá or Popayán, nor that the two neighbours reach Villavicencio over different roads, ducts, fibre owners or power domains.

The absence of a public interconnection profile makes the physical question harder. A request to the PeeringDB network interface for AS274248 returned no entity and an HTTP 404 response during review. That means no matching public network profile was available through that interface at that time. It does not mean Reintech lacks private interconnection, exchange access, cross-connects or additional arrangements. PeeringDB participation is voluntary, and small regional networks often interconnect without publishing facility detail there.

Even two genuine active sessions can fail together. Both can terminate on one Reintech border router. Both can enter the same building or roadside fibre. Both can rely on a shared long-haul owner between Villavicencio and another city. One neighbour can resell the other. A configuration error can be applied to both sessions. A power event can remove the router that hosts them. A route authorization problem can affect a prefix regardless of the number of sessions. Conversely, two sessions on separate routers, separate access providers and separate routes can materially improve recovery even when the external path view looks simple.

The title's distinction is therefore exact. Reintech now exposes two logical exits in strong current observations. It does not expose two proven physical paths out of Villavicencio. To establish the latter, a buyer or regulator would need at least the handoff locations, carriers or infrastructure owners, route drawings, shared-risk declarations, border-router separation, power arrangements and a test showing that traffic survives loss of each path. Public BGP is an invaluable external witness, but it sees routing decisions, not ducts and batteries.

The map stops at Villavicencio's address line

Reintech's geographic evidence has three different scales, and confusing them would overstate the company. The first is corporate geography: multiple records place the business in Villavicencio, Meta. The second is marketing geography: the company describes fibre service in the central urban area and radio coverage toward surrounding municipalities, rural settlements and farms. The third would be engineering geography: towers, fibre routes, aggregation sites, handoff buildings and serviceable polygons. The first two are visible; the third is not.

Villavicencio is therefore the defensible anchor, not a precise dot. The company and network registration records agree on the city. The two republished street addresses do not agree with each other, and neither is labelled as a network facility. A registered office can host administration while service equipment sits elsewhere. A rural address can be a commercial base without being a tower. An operational contact can work remotely. Turning any of these into a point of presence would collapse legal and physical geography into one unsupported claim.

The service descriptions are also deliberately broad. “Surrounding municipalities” does not name a municipality. “Veredas and farms” describes a customer type and settlement pattern, not a coverage contour. High-power radio can cross terrain where fibre construction is uneconomic, but line of sight, antenna height, vegetation, interference and rain margin decide whether a particular link is usable. No public page reviewed here provides tower coordinates, azimuths, channel assignments, Fresnel-clearance surveys, maximum link lengths or a serviceability map. A sales team may hold such information without publishing it.

Urban fibre language has similar limits. Saying that central Villavicencio has fibre-to-the-home service is not the same as publishing streets passed, premises connected, cabinet positions or route diversity. Fibre can run aerially on shared poles, underground in ducts, or through a mixture. It can use ring, tree or spur designs. Passive optical access commonly branches from one feeder toward many customers, but Reintech does not publish its splitter ratios or feeder layout. It would be speculative to draw a ring around the city or to assume that two nearby customers reach different aggregation points.

The address discrepancy is useful precisely because it blocks a false certainty. Carrera 5 Este 15-49 and Carrera 2 6 19, Vereda Barcelona, might both have had a legitimate relationship to the company at different times or for different purposes. The public material does not show whether either address carries active electronics. A serious location assessment would confirm current occupancy, inspect property and tower records where lawful, compare customer installation evidence, and obtain operator confirmation. This article does not convert commercial address listings into network facts.

External route data is not a substitute map. The registration addresses of AS262191 and AS273103 are in other Colombian cities, but traffic may interconnect at a neutral facility, a private handoff, a carrier node or a remote port elsewhere. An AS path carries autonomous-system numbers, not the GPS coordinates of each hop. Even an IP traceroute can be distorted by router addressing, label switching, private interfaces and return-path asymmetry. Any line drawn from Villavicencio to Bogotá or Popayán on the current evidence would be illustration, not a discovered route.

The honest map is thus a bounded statement: Reintech is centred in Villavicencio; it says it serves urban fibre customers and rural radio customers in and around that market; its external routing is visible through two neighbouring autonomous systems; exact access and exit paths remain undisclosed. This map is less visually satisfying than a line network, but more useful for risk. It tells customers what to request: a serviceability confirmation for the precise premise, the access medium, the first aggregation dependency, and whether a backup uses a genuinely separate local route.

It also frames regional importance. A provider does not need a national map to matter. A local network can be critical to homes and enterprises in places where national carriers have weak coverage or inflexible offers. Rural radio may reach customers whom fibre economics exclude. Local technicians may restore a last-mile problem faster than a distant support centre. The same concentration can increase correlated risk when many users depend on one city-level aggregation or one backhaul corridor. Reintech's geographic value and its geographic vulnerability can be two sides of the same local footprint.

Fibre, radio and television share a hidden aggregation problem

Reintech sells more than one kind of access, which broadens its commercial reach but complicates the dependency chain. Its structured-cabling page describes fibre, UTP, data-centre, AC and DC work, plus cabling for internet and television. This supports a picture of staff or contractors familiar with premises wiring and optical installation. It does not enumerate the company's deployed access plant or establish that every installation is built to the same standard.

Its remote-hands page goes further into active equipment. It names configuration work involving ONTs, OLTs and GPON, routing, switching, firewalls, routers and major equipment brands. Again, the page demonstrates an offered competence, not the count or ownership of Reintech devices. Still, these service descriptions help explain the likely operating surface. Fibre subscribers depend on customer optical terminals and an optical line terminal somewhere upstream; business customers may add managed routers or firewalls; field work and configuration authority become part of restoration.

Radio adds a different concentration point. A point-to-multipoint design can serve several customer radios from one central sector. That improves deployment economics in dispersed areas because the operator does not need to build fibre to every farm. It also means the central radio, mast, channel, backhaul and power supply may become shared dependencies. Point-to-point links can add backhaul capacity or connect sites, but one high-capacity radio is not necessarily an independent backup. Two radios mounted on the same structure and powered by the same DC system can disappear together.

The Reintech television page says the company integrates television through NUPLIN with TVYMAS. The commercial and technical arrangement is not fully described. Public material does not show whether all video reaches subscribers over the same IP access network, uses separate capacity, or relies on another delivery mechanism. The named integration does, however, show another partner boundary. Television quality may depend not only on Reintech's local access but also on content acquisition, platform availability, authentication and the TVYMAS relationship.

This is the hidden aggregation problem: diverse retail products can converge before they leave the region. Fibre, rural radio, managed networks and television may look independent at the subscriber edge while sharing an aggregation switch, router, address block, upstream session, power room, support desk or billing system. The current public evidence neither proves nor disproves that convergence. A customer buying a radio backup to a Reintech fibre line should therefore ask where the two paths first meet.

If both enter the same Reintech node and external route, the radio protects against a fibre drop but not against an aggregation or upstream failure.

The same question applies to business continuity. A branch office might buy a nominally dedicated circuit and a residential-style backup. If both are delivered by Reintech and terminate on the same OLT, tower, router or power system, the second contract may improve protection from some failures while leaving larger ones untouched. A better design might use a different access medium, different local provider, mobile connection or satellite path, but each has cost, latency and operational tradeoffs. What matters is not the label “backup”; it is the shared-risk boundary.

Aggregation also shapes fault isolation. When one household fails, the first suspicion may be a local cable, optical terminal, radio or power adapter. When a whole sector fails, the likely scope shifts toward a splitter branch, access switch, radio sector or power feed. When both fibre and radio users fail simultaneously, the investigation moves toward shared aggregation, external connectivity, name resolution or central power. Reintech does not publish a status page or topology that would let outsiders distinguish these cases quickly. Its support team may have internal telemetry, but customers see the outcome rather than the cause.

The product mix should thus be treated as evidence of operating breadth, not automatic resilience. It shows that Reintech works across physical and logical layers and can serve different local demand. It also increases the number of handoffs and partner dependencies that need clear ownership during recovery. The decisive questions are where traffic aggregates, how that layer is powered, which components are duplicated, and who has the access and authority to replace or reconfigure them.

Sold speed is not installed or survivable capacity

Reintech's plans page gives the clearest view of saleable capacity. During this review, residential offers were displayed at 200 Mbps for COP59,900 a month, 400 Mbps for COP79,900, 600 Mbps for COP99,900 and 900 Mbps for COP129,900. Business offers included 300 Mbps and 500 Mbps plans, plus a 1,000 Mbps dedicated offer displayed at COP399,900. The business language includes symmetric service, an IP address, guaranteed 1:1 capacity and a service commitment. These are current advertised offers, not measurements of what every installed line delivers.

Five capacity concepts need to remain separate. Allocated capacity is numbering space such as IPv4 and IPv6 prefixes. Installed capacity is the physical and electronic ability of fibres, radios, switches, OLTs, routers and external circuits. Saleable capacity is what the operator offers across plans. Usable capacity is what a customer can obtain at a particular time after contention, protocol overhead and faults. Survivable capacity is what remains after a component or path is lost. Reintech discloses meaningful evidence for allocated and saleable capacity; it discloses very little for installed, usable-at-peak or survivable capacity.

An advertised 900 Mbps residential port does not mean 900 Mbps is reserved end to end. Residential access networks usually depend on statistical sharing, but the sharing point and ratio vary. The public pages do not disclose how many customers share an optical branch, radio sector, aggregation uplink or external circuit. They do not publish peak-hour throughput distributions, latency, packet loss or the proportion of customers who can reach plan speed simultaneously. A headline rate can still be a fair retail offer; it simply cannot stand in for network-wide capacity.

The “1:1 guaranteed” business wording raises a more specific question. It could mean an uncontended access commitment, symmetric dedicated bandwidth, or a commercial ratio within a defined service segment. The page does not publish the contract terms needed to determine the exact boundary. Even a fully dedicated last-mile circuit can meet a shared router or external link. A buyer should ask for the committed information rate, measurement point, burst conditions, latency and loss objectives, maintenance exclusions, remedy, external-path treatment and whether the commitment survives a neighbour or equipment failure.

The company's claim of more than 50,000 active users cannot safely be combined with plan speeds to estimate traffic. “User” may mean subscribers, accounts, individual people, devices or a cumulative commercial measure. The mix across fibre, radio, television, residential and business products is not published. Multiplying the headline user count by a headline plan would produce an enormous but meaningless demand number. Real traffic depends on concurrency and usage, while required resilience depends on which customers need service at the same moment and which failures correlate them.

Wireless capacity is especially local. A radio's modulation and throughput depend on channel width, signal quality, interference, distance and weather conditions, while a sector is shared among connected stations. Reintech advertises spectrum analysis and high-capacity links, but no channel plan, sector load, licensed-band inventory, measured availability or backhaul size is public. Rural service may be valuable precisely because it works where fibre does not. Its usable capacity must still be tested at the customer location, especially during busy hours and adverse conditions.

Fibre has different unknowns. Public material does not list OLT ports, splitter ratios, feeder counts, headroom, optical budgets, route kilometres or aggregation uplinks. Nor does it reveal whether critical links form rings or end in spurs. A single high-capacity fibre can carry far more traffic than current subscribers require and still be a severe failure point if there is no alternate route. Conversely, a lower-capacity backup can preserve essential traffic if prioritisation and failover are engineered well. Capacity and redundancy are related, but neither implies the other.

External capacity is the largest blank. BGP shows logical adjacency to AS262191 and AS273103, but not contracted bandwidth, port speed, utilisation or failover headroom. One neighbour could carry most traffic while the other provides a limited path, or both could be comparably provisioned. One could be visible for only selected prefixes. The current path sample already shows asymmetry across the two IPv4 /24s. Without traffic data or circuit disclosure, it is impossible to determine whether the second neighbour can absorb the first neighbour's load.

Power capacity is also part of throughput. A router rated for gigabits is unavailable when its site loses power beyond battery runtime. A radio sector cannot serve customers after its DC system is exhausted. Reintech sells power-related services, examined below, but gives no inventory for its own sites. Spare equipment, replacement optics, trained staff and vehicle access are forms of recovery capacity too. The operational question is not only how many megabits the network can carry in normal conditions. It is how much service can be restored, for whom, and how quickly when normal conditions end.

IPv6 exists in routing before it appears in user measurements

Reintech's IPv6 position is simultaneously stronger and less complete than a route count suggests. The direct /32 registration and three active IPv6 announcements show deliberate network-level work. Route authorization is also visible. A RIPEstat RPKI check for 2803:77d0::/32 returned Valid for origin AS274248, with a covering authorization allowing announcements down to /48. That supports both the aggregate and the observed /48 more-specifics. Validity means the route origin and prefix length conform to the published authorization; it does not mean the route delivers working IPv6 to subscribers.

The two IPv4 routes illustrate the same distinction from another direction. A RPKI check for 38.211.254.0/24 returned Valid for AS274248 with maximum length /24. A check for 38.191.213.0/24 returned Unknown with no validating authorization in the response. Unknown is not Invalid. It means the relying party cannot validate the origin against a covering ROA. Given that the parent block is registered to TV&MAS, public clarification of the delegation and authorization would reduce uncertainty.

End-user observations remain thin. The APNIC IPv6 measurement table for AS274248 was last dated 13 July 2026 during review. Its latest raw interval showed nine observations and no IPv6-capable or IPv6-preferred samples. The 30-day interval showed about 32.46 observations and no measured capability. The 60-day interval showed about 47.96 observations with roughly 0.0379 per cent capability, and the 90-day interval about 63.13 observations with roughly 0.0186 per cent. Fractional counts reflect the statistical method. These are tiny samples, far too small to conclude that Reintech has no operational customer IPv6.

What the contrast does establish is a verification gap. A network can announce IPv6 at its border while residential routers, optical terminals, radio customer equipment, address assignment, DNS behaviour, firewall defaults or support processes remain IPv4-only. It can also provide IPv6 to a customer population that APNIC's advertising-based sample rarely reaches. The route is a prerequisite for broad native service, not proof of its penetration.

For a subscriber, operational IPv6 has several testable parts. The customer edge needs a global address or delegated prefix. Default routes and DNS need to work. Common destinations must be reachable without abnormal loss or delay. Firewalls must protect inbound traffic without breaking expected applications. Prefix delegation should survive routine reconnects or at least follow a documented stability policy. Support staff need to recognise failures that affect one address family. None of these details appears in Reintech's public plan descriptions.

For Reintech, the /32 is strategically useful even if current customer adoption is low. It provides room to assign cleanly structured /48s or smaller end-site prefixes, reduce dependence on scarce IPv4, and build an address plan that can grow. The two announced /48s may represent sites, policy groups, services or traffic-engineering choices, but their purpose is not disclosed. A broad allocation is an option set. Converting it into dependable customer service requires access equipment support, configuration, monitoring and trained support.

IPv6 also interacts with resilience. If IPv4 fails over between two neighbours but IPv6 does not, dual-stack applications may stall or select the degraded family. If the /48 more-specifics receive a different external policy from the /32, a neighbour loss may change reachability in ways that differ from IPv4. The current BGP sample sees both neighbours before the IPv6 routes, which is encouraging logical evidence. There is no public failover exercise showing how quickly each route converges or whether customer sessions recover.

The strongest fair statement is therefore limited: Reintech has a properly registered IPv6 /32, valid route-origin authorization for the aggregate through /48, and active global IPv6 announcements. Public user measurements do not yet provide a robust picture of adoption. Customers and partners should test native service directly rather than infer it from the allocation, and they should examine both normal reachability and loss of each external neighbour.

Power and field access decide whether a route remains usable

Every part of Reintech's apparent access mix depends on power, but not in the same place. A household optical terminal or radio normally depends on customer-premises electricity. An access point, OLT, aggregation switch and border router depend on operator-controlled or hosted power. A fibre span between passive components may need no mid-span electricity, while each radio relay does. Backup at one layer cannot compensate indefinitely for failure at another.

Reintech's renewable-energy service page advertises solar systems, batteries, UPS equipment, battery banks and DC backup in a stated range from 1K to 8K for critical telecommunications. This is relevant evidence of commercial familiarity with power continuity. It is not evidence that Reintech's own towers, access sites or routing equipment use those systems. The page does not list installed backup by location, runtime under current load, generator support, fuel arrangements, maintenance intervals or remote battery monitoring.

The distinction matters most at unmanned or hard-to-reach radio sites. A battery bank can bridge a short interruption, but its real runtime depends on load, age, temperature, charge state and maintenance. Solar can extend autonomy if generation and storage are sized for actual conditions. A UPS can protect against voltage disturbance without surviving a long outage. If a site also carries a high-capacity backhaul, losing it may affect far more users than losing one customer radio. None of those specific conditions can be assigned to Reintech sites from the sales page.

Field access is the next constraint. A technician may need permission to enter a roof, farm, tower compound, building room or third-party facility. Roads and weather can delay travel. Replacement radios need correct configuration and alignment. Fibre repair may require locating a break, securing a safe work area, obtaining splicing equipment and coordinating with pole, road or property authorities. A technically simple fault can therefore have a long elapsed recovery if access authority or transport is slow.

Villavicencio's municipal risk-management office description places telecommunications, transport and public services among the functions considered in emergency response. This is contextual evidence that connectivity sits within a broader local recovery system. It is not evidence of a Reintech incident, a formal priority arrangement or a particular local hazard affecting the company. It does underline that network restoration can depend on roads, electricity, public coordination and access beyond an operator's direct control.

Power failures also expose the difference between route visibility and service availability. An AS route can remain globally visible from an external session while a local access site is dark. If the border router loses power but a neighbour temporarily retains the last route, the announcement can even persist briefly despite failed forwarding. Conversely, a customer can lose premises power while Reintech's network remains fully available. Good incident analysis needs observations at several layers rather than a single global indicator.

The relevant questions are concrete. Which sites have batteries, for how long under tested load, and when were they last capacity-tested? Is there a generator or a safe connection for one? Are border routers and access aggregation on separate power feeds? Do two supposed paths share a utility transformer or building switchboard? Are spare radios, optics, power supplies and fibre materials held locally? Can staff reach critical locations outside ordinary hours? Does the monitoring system detect falling battery voltage before service stops?

Reintech advertises 24-hour support, but support availability is not the same as repair capacity. A support representative can receive a report at night while a field intervention waits for daylight, access approval or a partner. For rural users, remote diagnosis can distinguish a failed customer power supply from a sector problem, but physical replacement still takes travel. For business customers, a contractual escalation path should identify who can authorise rerouting, dispatch, equipment replacement and partner contact.

No public answer reviewed here supplies those details. That absence should not be read as proof of poor practice; many regional operators keep site security and operational plans private. It should be read as the boundary of the external assessment. Reintech's power-service expertise is a positive capability signal. Tested backup coverage across its own network remains unresolved, and so does the field-recovery depth behind the 99.9 per cent availability claim.

Failure starts with convergence and recovery starts with authority

A credible failure path begins with scope. Suppose a rural customer loses service. The first checks are premises power, cabling, radio status and whether nearby customers on the same sector are affected. If a whole sector is down, the investigation moves to the central radio, backhaul and site power. If fibre and radio areas fail together, the suspected boundary moves toward shared aggregation or external connectivity. If one address family fails while the other works, routing, authorization or customer-edge configuration becomes more likely. Each expansion in scope requires different evidence and a different person with authority to act.

One active measurement provides a narrow glimpse of the external path. The IPinfo page for 38.211.254.0/24 reported the prefix as RPKI-valid and displayed a ProbeNet traceroute made from Bogotá on 19 June 2026 that reached 38.211.254.1 through AS262191. The page also reported 11 ping-responsive addresses in its scan. One trace to one address at one time cannot map Reintech's physical route, customer population or failover. It does corroborate that AS262191 was carrying a reachable path toward one Reintech prefix from that vantage point.

If that path becomes impaired, BGP may withdraw a route or select another neighbour. Convergence time depends on session state, timers, policy, route propagation and whether the alternate path is valid and adequately provisioned. Existing customer sessions can still break as the path changes. Traffic may return asymmetrically. The second neighbour may see only some prefixes or have limited public evidence headroom. A route can converge before the local access fault is repaired, or the access can be intact while an external filtering error persists. “BGP recovered” and “customers recovered” are not interchangeable.

Route authorization adds another authority boundary. Reintech can publish routes from AS274248, but a parent-address holder may control the ROA or delegation for space registered elsewhere. Neighbouring networks control their filters and sessions. LACNIC controls resource-registration processes, not Reintech's live routers. When 38.191.213.0/24 is Unknown under RPKI, the immediate risk is not that the route must fail; many networks accept Unknown routes. The unresolved issue is who can create or adjust the appropriate authorization and how quickly they would coordinate during a routing-security event.

Customer authority begins with the operator's PQR page, which accepts petitions, complaints, claims and suggestions and frames them within response obligations. That is a formal consumer contact path, but urgent restoration also needs an operational escalation route. A PQR record can document harm and trigger a response; it does not by itself dispatch a technician or change a BGP policy. Business buyers should distinguish support tickets, network-operations escalation, contractual notices and regulatory complaints.

Colombia's compiled CRC Resolution 5050 provides a broader regulatory frame, including rules around service availability and automatic compensation for lack of internet or telephone service under applicable conditions. The exact remedy depends on service and circumstances, and a regulatory entitlement cannot restore packets in real time. It does create an incentive to record unavailability accurately and gives users a route beyond informal support.

The CRC Circular 149 of 2023 discusses availability measurement for fixed networks, including CMTS or OLT-based measurement and reporting expectations tied to subscriber scale. Its threshold treatment matters when interpreting silence: a smaller provider may still need to measure availability without appearing in the same public reporting set as an operator above the national subscriber threshold. Absence of a public metric is not proof that nothing is measured; it does leave outsiders unable to reconcile Reintech's 99.9 per cent claim with a defined regulatory series.

Recovery should proceed in an authority chain. The customer or monitoring system establishes scope and opens an operational record. Reintech identifies whether the fault is premises, access, aggregation, power, routing or partner-controlled. A field lead gains site access and replaces or repairs local equipment. A network lead changes routing or configuration where required. A resource holder or neighbour acts on authorization, filtering or external circuits. Customer service communicates what is known, avoids an unsupported restoration time, and confirms service at the user edge rather than only at the core.

The restoration test must match the failure. After a radio replacement, verify signal, loss, latency and throughput, not merely link lights. After fibre repair, check optical levels and affected branches. After neighbour failover, test all five announcements from multiple external vantage points and verify both IPv4 and IPv6 at customers. After a power event, test batteries under load and determine why monitoring or redundancy did not prevent the interruption. After a route-authorization change, confirm global validation status and propagation. A recovered dashboard without an end-user test can conceal a partial failure.

Reintech does not publish a detailed incident archive, independent uptime series, failover exercise or recovery-time distribution. That makes it impossible to grade actual restoration speed from the outside. The observed route history, current dual-neighbour view, formal complaint channel and regulatory context show pieces of a recovery environment. The missing piece is practiced evidence: named ownership, tested alternate paths, measured backup runtime, spare availability and a record showing how service returns after a real component loss.

What customers, partners and regulators should verify next

The assessment ends with a Medium network evidence grade. That is not a mediocre verdict. It reflects a meaningful positive base: REINTECH SAS has an operating public presence, consistent Colombian identity, service offers suited to Villavicencio's urban and rural market, its own autonomous system, a direct IPv6 allocation, one directly registered IPv4 /24, live dual-stack announcements and two external neighbours in strong current observations. A Negative grade would ignore this evidence. A Strong grade would pretend that route visibility answers questions it cannot answer.

For a household, the first verification is practical. Confirm whether the address receives fibre or radio, what equipment requires household power, who owns it, what installation work is included and what the support path is outside business hours. Ask whether a plan speed is an access maximum or a normal peak-hour expectation. If remote work, payments, security cameras or education depend on the connection, keep an alternate means of access that does not share the same local path. A second Reintech product may protect against one medium-specific fault, but its shared aggregation should be understood.

For a rural customer, radio-specific evidence matters. The installation should record line of sight, expected signal, mounting, grounding, cable route and power protection. The operator should explain whether the customer link reaches a sector or point-to-point node, what happens when interference rises, and whether the serving site has backup power. A spare customer radio in a warehouse is helpful only if staff can reach the premise and the central site remains working. Restoration commitments should account for travel and access, not only remote response.

For a business buying the 1 Gbps dedicated offer, the service definition should be written. The buyer needs the committed rate and measurement point, external-capacity treatment, latency and loss objectives, maintenance terms, remedy, address assignment, IPv6 availability and escalation authority. It should ask whether AS262191 and AS273103 are active for its specific prefix and site, whether they terminate on separate routers and power, and whether either path can carry the whole committed load. A live failover demonstration is more valuable than a diagram alone.

For a customer using public addresses, 38.191.213.0/24 deserves special clarification. The company can explain the delegation from the parent block's registrant, the expected stability of assignments and the path for establishing a validating ROA if that is contractually possible. Customers using allowlists, VPN peers or hosted services should know what happens if they must move to 38.211.254.0/24 or IPv6. Address portability and DNS-change time can dominate recovery even after physical connectivity returns.

For IPv6, the next evidence should come from customer edges. Reintech can publish whether residential and business products receive native IPv6, the delegated prefix size, router requirements and support status. A small set of distributed probes in urban fibre and rural radio areas could measure reachability, latency, loss and neighbour failover for both address families. That would turn a strong routing prerequisite into evidence of usable service and help explain the very small APNIC sample.

For partners, the central question is shared risk. AS262191, AS273103, TVYMAS and any access or facility counterparties know their own contractual and physical boundaries. Reintech can disclose enough to serious buyers without exposing sensitive coordinates: handoff city or facility class, distinct infrastructure owner, router separation, power domain, normal and emergency contact, and whether a route is primary, balanced or standby. A simple statement that two carriers exist is weaker than a table of which components remain after each simulated loss.

For regulators and local authorities, the important scale is affected users. The more than 50,000 active-user claim needs a definition before it can be used for public-risk planning. Subscriber counts by municipality, access medium and critical-service class would reveal concentration without identifying individuals. Availability measurement should distinguish planned maintenance, access faults, core faults and external events. Emergency coordination should identify which sites or routes serve health, public-safety, education or other essential users, while respecting security constraints.

For Reintech itself, the highest-value public disclosures would be modest. It need not reveal exact tower coordinates or commercially sensitive circuit prices. A dated network statement could identify active service municipalities, access media, dual-stack availability, the number of independently powered core sites, whether external paths are physically diverse, and the measurement basis for 99.9 per cent availability. A status and incident history could separate local access incidents from route events. A route-security statement could explain authorization for both IPv4 blocks.

Several unresolved facts should remain explicit until evidence changes them. No public record reviewed here establishes fibre-route kilometres, tower count, radio-channel inventory, OLT or splitter counts, border-router count, external circuit sizes, peak utilisation, generator coverage, battery runtime, spare stock, field-team depth, interconnection buildings, physical path diversity or tested recovery intervals. No public record maps products to the five routes. No robust end-user sample establishes IPv6 penetration. The second observed neighbour improves the topology picture but does not erase any of these gaps.

That unresolved list is not an accusation. Regional operators often create value with lean teams and partner infrastructure, and detailed disclosure has security and commercial costs. Fact-versus-inference discipline protects Reintech as much as it protects customers: it prevents a broad allocation from being mistaken for huge installed capacity, but it also prevents a missing PeeringDB profile or sparse measurement sample from being mistaken for absence of service. The evidence should be allowed to say exactly what it says.

Reintech's significance lies in the physical middle. It connects Villavicencio customers who may sit beyond easy national-carrier economics; it links fibre, radio and partner services to the global routing system; and it now shows two logical external neighbours. The next standard of proof is not another route count.

It is whether one failed fibre, radio site, power domain, router or neighbour can be removed while affected customers keep a usable service—and whether the people with the authority, access and spare equipment to restore the failed component can do so before local dependence turns a compact network fault into a regional disruption.