Summary
- Rentic's public record supports a narrow but credible operating picture: the Quibdó company says it began serving homes and businesses by radio link in 2012, later developed fibre access, documented an FTTH footprint in a 2024 public competition, registered AS273008 and now originates one IPv4 /23 and one IPv6 /48. None of those records discloses subscriber count, busy-hour traffic, upstream bandwidth, optical split, cabinet elevation, battery runtime or the share of premises that can receive service today.
- The distinction between installed and usable capacity is unusually important in Quibdó. Municipal and environmental records describe recurrent flooding on the Atrato alluvial plain, rain-triggered landslides and exposed public infrastructure; the regional electricity distributor records weather, vegetation, blocked roads and river conditions as causes of faults or delayed repair. A prefix can remain visible at distant route collectors while a flooded access enclosure, failed pole supply, dark customer terminal or unreachable splice removes service from part of the city.
- Public route views show two immediate neighbours for both originated prefixes, AS18678 and AS262589, but they do not reveal two contracts, two ports, two physical paths or independent power. Chocó's documented transport failures in 2025 show why that boundary matters: fibre damage on named regional corridors disrupted voice and data and delayed repair in difficult terrain. Rentic's resilience can therefore be judged only after its logical diversity is tied to physical routes, hand-offs, power domains, tested failover, local spares and restoration authority.
The route can remain visible while the street goes dark
After intense rain, imagine a technician standing beside a low access cabinet in a Quibdó neighbourhood. Shallow brown water covers the edge of the road. A fibre lead enters from a pole, a wireless mounting point is visible above the roofs, and a portable meter stays dry under the technician's rain cape. The technician is not at a documented Rentic site; the scene is an analytical reconstruction of the physical work implied by a fixed-and-wireless operator in this city. Its purpose is to follow one customer's service in the opposite direction from the way the public internet sees it.
The customer's path starts in a room, shop or school, not in a routing table. A powered terminal must convert light or radio into an Ethernet or Wi-Fi connection. The drop must reach a splitter, cabinet or radio sector. Any active access equipment needs electricity and a functioning enclosure. Aggregation must carry the session to a hand-off where Rentic can reach one or more wider networks. Only after those local conditions hold does an advertised address become useful to the person paying for it.
A remote observer can still see AS273008 originate a route when a subset of drops, a cabinet, a pole transformer or an entire access ring has failed.
That separation is not an exotic edge case in Quibdó. The municipal government reported severe flooding after persistent rain in August 2024, with the heaviest immediate effects on rural communities. The notice does not name Rentic, a communications cabinet or a fibre route. It does establish that water can become an operating condition rather than a remote design assumption. A local outage after such rain could arise from direct water ingress, a safety shutdown, damaged overhead plant, loss of commercial power, a dark customer terminal, or the inability of a crew to reach the failed element. Those mechanisms have different remedies even when the symptom is identical.
Quibdó's municipal disaster-risk plan gives the physical context more precision. It treats flooding on the Atrato's alluvial plain and terraces as recurrent, identifies intense rain as a principal cause, and notes that affected collective assets can include roads and public-service infrastructure. It also describes rain-saturated soils as a trigger for slides in low hills and lists telecommunications antennas among assets requiring exposure assessment. The plan is older planning evidence, not a current Rentic asset inventory. Its value is to show that “Quibdó” contains several hazard surfaces: low floodplain, obstructed drainage, unstable slopes, urban roads and dispersed rural approaches.
At the cabinet, resilience begins with choices that cannot be read from an IP address. Is the enclosure above an observed flood level? Are cable entries sealed while allowing drainage and ventilation? Is the fibre slack stored where water and vehicles cannot damage it? Does a radio site share the same pole and power feed as the fibre node intended to back it up? Can a technician isolate electricity before opening wet plant? Are optical closures, connectors, power supplies and customer terminals stocked locally, or must replacements cross a mountain road after a failure? A route announcement answers none of these questions.
This is why the apparent paradox is straightforward. Internet reachability is an end-to-end property assembled from layers that fail separately. The route can remain visible because a core router and one upstream session remain alive, while homes below a damaged splitter disappear. The route can also vanish while a local access network remains powered but cannot reach the wider internet. Good recovery practice begins by locating the break before interpreting a green routing indicator as evidence that the street is online.
One Quibdó company sits behind several dated identities
RENTIC S.A.S. is the relevant legal name, but its public records describe different aspects of the company at different dates. That matters because an operator, an address holder, an applicant for public support and a customer-facing brand can overlap without being interchangeable. A sound assessment must keep the legal entity, access operation, public routing identity and third-party infrastructure boundaries separate.
The company's current service-channel terms identify RENTIC S.A.S. at Calle 18 #21-46 in Quibdó. They say WhatsApp is used for customer care, support and billing notifications during weekday office hours, while expressly declining to guarantee immediate replies or uninterrupted availability of the messaging service. This is evidence of a customer-contact surface, not a network service-level commitment. It does not say that internet faults wait for office hours, nor does it disclose a network operations centre, after-hours escalation, restoration target or emergency number.
The related privacy notice says the company may handle a contract number, service address, billing information and support messages. Those categories are consistent with an operator maintaining individual service relationships. They still do not reveal how many active contracts exist, whether each is fixed fibre or wireless, where customers are concentrated, or which company owns the drop and active equipment. Even a service address is a commercial record until paired with a verified serviceability and asset map.
A more formal snapshot appears in the Ministry of Information and Communications Technologies' final evaluation of a 2024 fixed-broadband competition. It names RENTIC S.A.S. under tax identifier 900999224-5 and records the applicant as then registered with the ministry as a current value-added internet service provider. It also says the company had submitted the fourth-quarter 2023 regulatory report. Those findings are bounded by the competition's June 2024 evaluation date; they are not a perpetual status certificate.
The same evaluation is important for what went wrong. Rentic did not pass the legal eligibility stage because the ministry recorded an unpaid amount for 2019, while noting that later years had been paid normally. That is a historical finding about one competition. It should not be turned into a claim that the company is currently in arrears, has lost authority to operate, or has ceased serving customers. The evaluation also shows that Rentic's technical proposal for Quibdó was considered valid after clarifications, yet the company was not selected because it was not legally enabled for the award.
The boundary has practical consequences. The competition referred to an Internexa node and required participating providers to connect new low-income households. A project applicant might own local fibre while purchasing transport at a designated hand-off; it might lease ducts, poles, sites or capacity from others; it might install customer terminals that remain company property. The public evaluation establishes neither the final ownership chain nor any completed project under that award. Rentic's proposed target cannot be added to its customer base because the award did not go to Rentic.
Rentic's own history moves from radio to fibre
Rentic's public home page supplies the clearest company account of how its access business developed. It says the enterprise was created in 2012 after an earlier internet-room business, and that founder Bayron Enrique Mena began providing broadband by radio link to homes and companies in Quibdó. It describes Rentic by 2020 as strong in broadband over fibre and says the company also opened technical network-support services to telecommunications businesses. The page markets high-speed fibre, education and information-technology projects.
This account supports a mixed access history, but not a current technology split. A radio-link origin does not prove that a fixed-wireless network still carries customers in 2026. A fibre offer does not prove that every service area has fibre to the premises. Nor does the phrase “high speed” define a plan rate, contention ratio, optical budget or measured experience. The company has chosen to tell a progression from radio to fibre; outsiders still need current plant and service records to quantify what survives from each stage.
The page's public content metadata dates the original home-page entry to February 2023 and records a modification on 14 July 2026. That recent modification is useful evidence that the page was maintained close to this assessment date. It does not date every sentence or establish when a pictured or described asset entered service. In particular, the history's reference to the company's position in 2020 should remain a statement about that year, not a claim that its present fibre footprint is unchanged.
The 2024 ministry evaluation gives the fibre claim a second, more technical basis. Rentic's submission passed checks requiring a percentage of homes passed by an already deployed fibre network in Quibdó, an FTTH topology file identifying premises passed but not connected, a count of current connected homes, specifications for customer optical terminals, and a weekly installation plan. The public report does not reproduce Rentic's topology, subscriber count or percentage. It does show that evaluators received enough material to mark those items compliant.
One number in the final table demands special discipline: Rentic proposed connecting 20,000 new households at the competition's specified 25 Mbps downstream and 5 Mbps upstream service. Its technical proposal was marked valid for that quantity. This was an application for a subsidised expansion, not a record of 20,000 existing subscribers, premises already connected, installed terminal ports or completed installations. Because Rentic failed the legal eligibility test, the table assigns it no households.
Treating the proposed number as operational scale would erase the difference between an ambition, a technically reviewed plan and an awarded build.
The underlying topology remains unavailable to the public. Without it, there is no reliable route from the word “fibre” to a street-level footprint. The network could have long feeder runs and sparse splits, compact urban clusters, radio extensions from fibre-fed sites, or several generations of plant with different maintenance needs. No public map gives feeder kilometres, pole ownership, underground segments, splice locations, splitter ratios, radio frequencies, sector loading or premises passed.
The physical operating surface is real enough to have been documented to evaluators, but its size and condition cannot be audited from the released report.
Two live prefixes are not a capacity figure
At the internet layer, Rentic's identity is unambiguous. LACNIC's record for AS273008 shows a direct autonomous-system allocation registered on 23 May 2023 to RENTIC S.A.S. in Quibdó. An autonomous system gives the company a number under which it can originate routes and apply routing policy. It says nothing about the amount of traffic those routes can carry, the number of upstream ports, or the local access plant behind the router.
The address holdings are asymmetric. LACNIC's IPv6 registration assigns Rentic the full 2803:b590::/32, registered in April 2023. ARIN's IPv4 record describes 38.253.68.0 through 38.253.69.255, equivalent to a /23, as an assignment named RENTIC-CGNT-NET-1 beneath a larger Cogent block. The careful language is therefore that Rentic has an IPv6 allocation and an IPv4 assignment for use. Neither address quantity should be converted into subscribers, ports or throughput.
A current RIPEstat announced-prefix view, observed on 17 July 2026, lists exactly two originated prefixes: 38.253.68.0/23 and 2803:b590:1::/48. The IPv4 announcement uses the whole registered /23. The IPv6 announcement uses one /48 from the much larger /32. That is a large difference in address space but no basis for saying that IPv6 has more usable bandwidth. Address plans and forwarding capacity are independent.
The IPv4 routing-status record dates the route's first observation with origin AS273008 to 26 October 2023 and saw it from 324 of 327 relevant RIPE RIS peers on 17 July 2026. The IPv6 routing-status record dates first observation to 24 January 2024 and saw it from all 322 relevant IPv6 peers in that view. These counts indicate broad collector visibility at a moment in time. They are not customers, routes in Rentic's own routers, packets per second or a service-availability percentage.
Even the security state differs by family. The route-origin validation response for the IPv4 /23 returned “unknown”, meaning no validating route-origin authorisation was found for that origin and prefix in the consulted view. Unknown is not invalid; it means relying networks cannot validate this announcement through that mechanism. The same validation response for the announced IPv6 /48 returned valid for an exact authorisation to AS273008. This improves origin validation for the visible IPv6 route, but it does not protect a fibre from floodwater or keep an access node powered.
Installed capacity must therefore be described in several units. Address inventory includes the IPv4 /23 and IPv6 /32. Publicly originated address space includes the IPv4 /23 and one IPv6 /48. Documented access installation includes some FTTH network sufficient for the 2024 proposal's required topology and coverage declarations, but the released evaluation omits quantities. Proposed service scale included 20,000 new households at 25/5 Mbps under a competition Rentic did not win.
Usable capacity would require current upstream bandwidth, aggregation limits, optical split and load, radio spectrum and sector utilisation, access-port occupancy, power endurance and busy-hour traffic. None is public.
This layered account may sound less decisive than a single capacity number, but it is more useful. An address block can be almost empty. A 10-gigabit port can be constrained by a smaller commercial commitment. A fibre plant can pass a house without a drop, terminal or affordable subscription. A customer can have a working terminal but no power. Two healthy routes can share one congested hand-off. Capacity becomes service only when every segment can carry the offered load at the time users need it.
Two visible neighbours do not prove two physical exits
The current AS-neighbour view for AS273008 identifies two immediate provider-side neighbours: AS18678 and AS262589. That is stronger evidence than a marketing claim of redundancy because it reflects observed routing paths. It is still logical evidence. The view does not expose contract terms, ports, hand-off buildings, fibre owners, route geography, power feeds or whether one session is primary and the other rarely used.
The detailed IPv4 route-collector state contained 332 observed paths at retrieval, all ending in AS273008, with either AS18678 or AS262589 immediately before the origin. The corresponding IPv6 route-collector state contained 341 paths and the same two immediate predecessors. These are multiple observations of propagation through collectors, not hundreds of Rentic circuits. Their value is to show that both address families can be seen through both neighbouring autonomous systems in the public view.
Registry identity helps label the neighbours without inventing relationships. LACNIC identifies AS18678 with INTERNEXA S.A. E.S.P. LACNIC identifies AS262589 with SAMM TECNOLOGIA E TELECOMUNICACOES S.A. Those registrations do not state that Rentic has purchased two independent services, or that either company owns the final fibre into Rentic. A neighbour can appear through transit, peering, a reseller arrangement or another commercial structure. The public routing view establishes adjacency, not the contract.
Physical diversity requires more demanding questions. Do the two sessions terminate on separate Rentic routers? Are the routers in different rooms or sites? Do the hand-offs leave Quibdó over different cables and corridors? Do they traverse different carrier nodes before converging? Are the optics, patch panels, poles and building entries independent? Can each path carry the normal busy-hour load on its own, or would failover create congestion? Do both routers, hand-offs and access aggregators survive the same utility interruption? Has traffic actually shifted during a controlled test?
The answer could be better or worse than the public diagram implied by two neighbours. Two sessions might travel over genuinely separate routes and power domains, making the observable diversity meaningful. They might also share one local lateral before separating, leaving a single excavator, pole failure, flood zone or equipment room able to remove both. Conversely, one neighbour could deliver a protected service over two carrier paths that the public route view collapses into one adjacency. Autonomous-system count is therefore neither an upper nor a lower bound on physical route count.
Routing policy also determines whether diversity is usable. Rentic needs filters, accepted prefix limits, local preferences, route-origin controls and a failover arrangement that does not strand one address family. The fact that both predecessors appear for both prefixes is encouraging, but it does not show normal traffic share or spare headroom. If one upstream carries nearly all traffic and the other is sized only for emergency control, the second session may preserve reachability while customer performance collapses. If return paths behave differently, a partial failure can be harder to diagnose than a clean withdrawal.
The public record therefore supports a precise conclusion: AS273008 has two observed immediate neighbours for both live prefixes as of the retrieval date. It does not support the phrases “dual-homed with full redundancy”, “two independent fibre routes” or “automatic failover with full capacity”. Those stronger claims need carrier letters, demarcation records, route drawings, port capacities, power diagrams and a recent failover result. Until then, the two neighbours are evidence of logical options, not proof that floodwater cannot find their common point.
The access network meets a city organised by water
Quibdó's geography is not merely background scenery for an internet provider. It decides where a cabinet can stand, which pole line stays reachable, how a fibre route crosses drainage, whether a buried closure sits below water, and how quickly a technician can move between urban and rural service areas. The city lies within the Atrato river system, where water is both a transport medium and a recurring hazard.
CODECHOCÓ's red alert for Quibdó, Medio Atrato and Medio Baudó drew on hydrological stations and observed Atrato levels at Quibdó's riverfront. It highlighted settlements on river margins and alluvial plains. The alert dates from July 2022 and cannot locate a present-day Rentic asset. It does show that flood exposure can be monitored and geographically differentiated rather than treated as an undifferentiated annual rain problem.
The slope side matters too. The national risk information system's high-landslide-threat alert for Quibdó, updated in June 2026, links the warning to intense or prolonged rain and terrain susceptibility. A landslide need not strike an access cabinet to affect service. It can block a road to a splice, damage a pole route, interrupt commercial power, prevent delivery of fuel or replacement equipment, or isolate a crew from a rural site. Recovery time is therefore partly a mobility quantity.
The national government's 2024 disaster declaration records 18 climate events in Chocó from 10 September to 12 November, with flooding accounting for 72.2 per cent, windstorms 16.7 per cent, landslides 5.6 per cent and wildfire 5.6 per cent. It cites Quibdó's municipal calamity declaration among those in the department. These are department-wide emergency statistics, not Rentic incident counts. Their relevance is to planning: several hazards can affect the same repair chain in one season.
A physical service map should reflect these differences. It would place headend and aggregation sites, active and passive cabinets, fibre feeders, radio sites, river and bridge crossings, pole routes, carrier demarcations, commercial-power feeds, flood levels, slope exposure, vehicle access, boat access and local spare stores. It would distinguish a route drawn along a road from one actually attached to roadside poles. It would also mark where two supposed exits share a conduit, bridge, tower, substation or equipment room.
No such Rentic map is public. The topology submitted in 2024 appears to have been detailed enough to mark FTTH premises passed but not connected, yet the evaluation releases only pass/fail findings. That creates an unusual evidence boundary: the existence of a map is supported, but its geometry is not. It would be wrong to reconstruct street routes from office addresses, customer contact details or route announcements. Even the low cabinet in the opening remains an illustrative risk location rather than a claim about a named Rentic enclosure.
Every advertised megabit borrows from the power system
The internet path depends on electricity at several points even when most of the outside fibre is passive. Customer terminals and Wi-Fi routers need household or business power. Optical-line terminals, switches, routers, radio equipment and monitoring systems need site power. Cooling, lighting and security can matter during an extended interruption. Batteries and generators only change the time at which a power failure becomes a communications failure; they do not remove the dependency.
The local distribution context is visible through DISPAC's public site, which identifies the company as the electricity distributor and retailer serving its Chocó market from Quibdó. This does not prove that every Rentic site buys electricity from DISPAC, identify a feeder, or establish a shared contract. It does establish the relevant operating environment in which a powered communications node or customer premises in Quibdó would normally sit.
DISPAC's 2024 management report is unusually candid about that environment. It reports 59,599 electricity customers in Quibdó at year-end, 95 per cent classed as urban, and describes the city's concentration as 55 per cent of its entire customer base. Those figures are electricity accounts, not Rentic customers or premises passed. They show the scale of the local power surface to which communications demand is attached.
The report gives reliability measurements at utility level: a 2024 SAIDI of 41.22 hours and SAIFI of 14.41, both within the applicable regulatory limits. SAIDI and SAIFI are aggregate distribution indicators, not a prediction for a particular Rentic address. A site on one feeder can perform better or worse than the system average. Yet the figures make an essential point: commercial-power interruptions are measurable operating events, not a hypothetical footnote. A cabinet with four hours of battery cannot be described as resilient to a year in which some interruptions exceed that duration without site-specific evidence.
DISPAC describes precipitation, saturated soil, falling trees, slides, blocked roads, wind, public-order constraints and difficult river access as causes of faults or delayed maintenance. It says flood conditions sometimes required supply suspension for public safety. The report also notes low-voltage risks under certain transmission contingencies and planned reinforcement around Quibdó. These are electricity-system observations. They should not be misreported as a history of Rentic outages, but they identify common-cause mechanisms capable of affecting access equipment, customers and repair crews together.
Power diversity must be tested along the entire service chain. A generator at a core room does not keep an unprotected neighbourhood switch alive. A battery in a radio cabinet does not power the customer's receiver. Two upstream hand-offs in one room still share the room's transfer switch, fuel and cooling. A wireless fallback mounted on the same pole as the damaged fibre can fail from the same tree or power event. Conversely, a passive splitter above flood level may need no local electricity at all, shifting the critical power points to the central terminal and customer premises.
The missing Rentic evidence is specific: no public record gives site power architecture, battery age, tested runtime under load, generator capacity, refuelling plan, low-voltage protection, surge protection, remote alarms, automatic transfer behaviour or priority sites. Nor is there a published list separating passive closures from active cabinets. Without those facts, “backup power” would be an empty phrase even if a battery photograph existed. Usable endurance is the shortest supported duration across all powered elements needed for a customer session.
Recovery authority matters as much as equipment. A telecommunications crew cannot safely repair an energised cabinet standing in water. It may need the electricity distributor to isolate a circuit, a pole owner to grant access, a carrier to test a hand-off, or municipal responders to clear a route. Contact trees, authorisations and safe-work rules can save more time than an extra spare kept in the wrong place. In a flood-prone city, power and access coordination are part of network capacity because they decide how long installed equipment remains useful.
The first users cut off are more than an address count
The cost of a local failure cannot be inferred from Rentic's address blocks. IPv4 sharing can place many households behind one public address, while IPv6 can assign many addresses to one customer. Some routed addresses may support servers, management equipment or point-to-point links. The number of affected people depends on the failed physical segment, time of day and the services those people are trying to reach.
DANE's 2018 census profile for Quibdó counted 120,679 residents and adjusted the municipal population to 129,237, with 36,380 private households. It reported internet access at 28.4 per cent of occupied dwellings with people present, subject to the profile's response notes. These are historical municipal figures, not Rentic's market share. They do show why one street cabinet can represent more than a handful of devices and why the unconnected population cannot be counted as latent Rentic demand.
The digital gap remained visible in DANE's 2024 quality-of-life survey. Nationally, 65.6 per cent of households had a fixed or mobile internet connection; Chocó was one of three departments below 30 per cent. This is a department estimate covering any internet method, not a fixed-broadband penetration figure for Quibdó. It establishes that loss of one working local connection can matter in a setting where alternatives are not universal.
Government expansion also changes the market without revealing Rentic's share. In April 2026 the presidency reported more than 12,000 Chocó households connected under national programmes, alongside more than 400 digital centres and 56 community peace-zone points in operation. The announcement does not name Rentic as an operator or beneficiary. It demonstrates that household, school and community connectivity increasingly coexist, so an outage may affect users through several institutional arrangements.
The first users harmed by a failure are often those beyond the least protected branch: customers behind one radio sector, one low splitter, one damaged pole span or one unpowered access node. A shop may lose electronic payments. A household may lose work or study access. A school or community point may concentrate many users behind one link. A support channel that relies on the same failed internet or on unavailable mobile coverage can make the outage harder to report. These are plausible impact mechanisms, not evidence that any named Rentic customer experienced them.
No reliable Rentic subscriber number is public. The ministry proposal's 20,000 target cannot fill that gap, nor can electricity customers, census households or all users seen behind an ASN. Revenue, penetration, churn and average traffic therefore remain unquantified. The defensible impact statement is narrower: Rentic presents itself as serving homes and businesses in Quibdó, and public evidence confirms a fibre access footprint and active internet routes. The number and type of users behind each failure domain remain unresolved.
A local fault can become a regional recovery problem
Chocó's recent transport incidents show how a fault outside an access provider's plant can overwhelm local redundancy. They do not establish that Rentic used the affected network, lost service, or shared the named route. They are valuable because they document real mechanisms—electrical damage, burnt or cut fibre, difficult terrain, high structures and repeat faults—on the wider connectivity surface around Quibdó.
On 12 April 2025, the ministry reported a cut on Azteca Comunicaciones' transport trunk, attributed at the time to electrical damage. Claro's information put Condoto, Río Quito and Nóvita in total suspension and Quibdó at 62 per cent affected. The ministry said people, commerce and financial services were harmed, while field work was delayed by difficult terrain and the height of the structures carrying the fibre. This was a regional incident reported through other operators, not a Rentic outage report.
Eight days later, a ministry follow-up on a new Chocó failure cited two fibre damages near Santa Cecilia, on the Tadó–San José and Tadó–Santa Cecilia sections. It said crews found cuts at two points and that service was being restored while the company investigated possible vandalism. The notice also referred to the same sector as the earlier incident and another event six months before. Repetition on a corridor is exactly why route labels must be tied to physical common points.
The concentration problem had already been acknowledged publicly. In April 2024 the ministry announced a second connectivity line for Chocó, saying the then-existing Azteca line could leave the department without connectivity when damaged. The government said it had contracted Internexa for a second channel, intended initially to support 23,100 households, and also discussed more mobile base stations and a programme for small and medium internet providers. An announcement of construction and expected benefit is not proof that a protected second route is complete, commissioned or used by Rentic.
Road access can turn a communications repair into a logistics problem. Invías reported the total closure of the Quibdó–Pereira corridor after a large landslide between Santa Cecilia and Pueblo Rico in December 2022. That event predates Rentic's ASN and says nothing about a Rentic route. It demonstrates that the same broad corridor named in later fibre incidents can become impassable, affecting arrival of crews, spares or fuel even when the communications cable fails elsewhere.
For Rentic, the recovery path should be tested as a sequence of authority. Customer reports must be grouped into a likely failure domain. Local monitoring should distinguish access loss, power loss, aggregation failure and upstream-session loss. A field lead decides whether dispatch is safe and reachable. The responsible carrier tests the demarcation when the fault crosses ownership. If one upstream remains, traffic policy must shift without exceeding its usable limit. Communications staff need an estimate grounded in access and carrier information rather than the continued visibility of a prefix.
The hardest cases are partial failures. A route may remain up through one neighbour while congestion or asymmetric paths impair applications. One neighbourhood may be dark while the rest of Quibdó works. Commercial power may return before wet equipment is safe to energise. A crew may repair fibre but wait for a carrier port or replacement power supply. Recovery is complete only when customer sessions and service quality return, not when a route reappears or an upstream ticket closes.
Public records disclose none of Rentic's restoration authority, escalation contacts, carrier repair priority, mean repair time, spare locations or tested alternate route. The regional incidents show why those facts matter without proving that Rentic shares the failures. A small operator can be locally competent and still wait on a road, utility, pole owner or carrier. Resilience is partly the ability to shorten those waits before the rain begins.
Resilience has to be built below the routing table
The most effective improvements follow the failure chain rather than the visibility of the brand. First comes an authoritative asset and dependency record. Every active node should be tied to its customers or downstream plant, power source, backup endurance, carrier hand-off, access route, flood and slope exposure, and compatible spare. Passive closures should be separated from powered cabinets so staff do not assume every enclosure has the same safety and recovery needs.
Second comes physical protection. Low equipment can be elevated above locally observed flood levels with sealed entries, strain relief, drainage and safe isolation. Fibre slack and closures can be kept away from vehicle paths and predictable water collection. Pole and radio assets can be inspected for vegetation and wind exposure. A second access method should not be placed on the same vulnerable pole, feed or room if it is intended to protect against those failures. None of these measures can be claimed as present at Rentic without site evidence.
Third comes power endurance expressed in time, not in the mere presence of batteries. Each required active element needs a tested load and runtime. Batteries need age and replacement records. Generator plans need fuel, access and safe operation. The customer edge also needs attention: keeping a central router alive does not help a household whose optical terminal is dark. Critical institutional customers may need coordinated premises backup, while ordinary customers need honest guidance about which equipment must be powered.
Fourth comes usable upstream diversity. The two observed neighbours offer a starting point for verification. Rentic could document distinct demarcations and paths, identify common ducts or carrier nodes, measure normal and failover load, and test each address family. A controlled test should prove that both IPv4 and IPv6 traffic move as intended, monitoring remains reachable, and the surviving service has enough headroom. A second session that only keeps the route visible is useful for control but should not be described as full service redundancy.
Fifth comes local repair capacity. Flood and landslide exposure makes the location of spares strategic. Compatible optical modules, splitters, closures, drop cable, connectors, power supplies, radio units and customer terminals are valuable only if they are dry, catalogued and reachable. Crew capability should cover fibre, radio, power safety and carrier escalation without depending on one person's memory. Alternate access by vehicle or boat may be necessary for some sites, but only an actual service map can show where.
Finally comes customer-centred restoration. Alarms should map to affected areas and priority services. Support staff need a channel independent enough to remain available during a fixed-network failure. Updates should distinguish diagnosis, access delay, repair, testing and restored service. Post-incident review should compare route state with access alarms and customer experience, because a stable prefix can conceal a neighbourhood outage. Measured restoration of sessions is a stronger completion test than a green routing session.
These measures also clarify economics. Elevating every enclosure, duplicating every feeder and maintaining unlimited spares would be costly. Risk-based investment should identify the branches with many users, critical services, repeated water exposure, long travel time or shared upstream dependencies. Public support may help expansion, but the 2024 competition shows that legal and financial eligibility can decide whether a technically valid plan receives an award. Operational resilience includes the institutional capacity to qualify for, execute and sustain investment.
The decisive evidence remains outside public view
Rentic can now be described with more precision than a generic label such as “regional ISP”. It is a Quibdó company that traces its broadband activity to 2012, describes a move from radio links toward fibre, documented an existing FTTH footprint for evaluators in 2024, holds an autonomous system and IPv6 allocation, uses an assigned IPv4 /23, and currently originates an IPv4 /23 plus an IPv6 /48. Both routes are visible through two immediate neighbouring autonomous systems. This is a credible operating and routing identity.
The same record leaves the service-delivery question open. There is no public subscriber count, current premises-passed figure, serviceability map, fibre length, optical-line terminal inventory, splitter ratio, radio-site list, spectrum detail, address-family adoption rate, busy-hour traffic, transit commitment, port speed, congestion record, latency record, power architecture, battery endurance, generator plan, spare inventory, crew roster, restoration history or tested failover result. These are not decorative details. Together they determine installed, usable and recoverable capacity.
Several documents could resolve the uncertainty without exposing customer data or security-sensitive detail. A redacted physical-diversity statement could show whether the two upstream sessions use separate routes, sites and power. Aggregate capacity records could state port speeds, normal peak load and failover headroom. A service map could publish neighbourhood-level coverage while withholding exact equipment coordinates. A power register could give tested endurance by node class. Incident summaries could report cause, affected service area and restoration duration without naming customers.
The most important disclosure would bind layers together. For each major failure domain, Rentic could identify the number or category of downstream users, local access technology, upstream path, power class, safe access method, responsible repair party and target restoration sequence. That would show whether a route seen by the world corresponds to a service that can be restored in the street. It would also reveal where one intervention—raising a cabinet, relocating spares, separating a hand-off or increasing battery runtime—removes several risks at once.
Until that evidence appears, fact and inference must stay separate. It is a fact that Rentic's two prefixes were visible on 17 July 2026. It is a fact that two immediate routing neighbours appeared for both families. It is a fact that a 2024 technical proposal contained FTTH coverage and topology material and proposed 20,000 new connections, and that Rentic received no household award in the final evaluation. It is a fact that Quibdó and Chocó have documented flood, power, transport and access hazards. It is an inference that Rentic plant may encounter some of those hazards; the location and severity for any particular asset are unknown.
That restraint does not weaken the resilience finding. It sharpens it. The visible internet ends at an origin announcement, while usable access continues through carrier ports, powered rooms, aggregation, fibre or radio plant, customer drops and premises electricity. Failure can enter at any point. Recovery can stall wherever ownership, safety, access or spare equipment changes hands. Quibdó's water and terrain make those hidden links operationally decisive.
The image of a technician beside shallow floodwater should therefore be read as a test question, not as a claim of a past incident. Which customers sit beyond this enclosure? Is it passive or powered? How high can water rise? Where does its feeder go? Which upstream remains if that feeder fails? How long does backup last? Who can reach the site, isolate it, repair it and verify customers are back? AS273008 cannot answer. Rentic's resilience lies in whether the company can.