CORPORACIÓN LORIATELECOM S.A.S. sells 800 Mbps FTTH in Cotopaxi, but its coverage polygon shows no routes or recovery design

Summary

• CORPORACIÓN LORIATELECOM S.A.S., trading publicly as Fiber Spot, offers monthly prepaid FTTH plans from 200 Mbps to 800 Mbps and names offices in Saquisilí, Mulaló, Pujilí and Latacunga.
• Its current public routing footprint is real: AS273035 originates an active IPv4 /23 and IPv6 /32, with meaningful routes visible through UFINET and AS263238, but logical diversity does not prove physically separate fibres, handoff sites or power systems.
• The decisive resilience facts remain undisclosed: access routes, OLT and splitter loading, upstream port sizes, peak utilisation, cabinet and site backup power, spare equipment, field-crew coverage and the number of customers concentrated behind each aggregation point.

At sunset, the relay is a question

At sunset above a Cotopaxi valley, a hilltop radio handing traffic into a fibre cabinet would be an intuitive picture of a mountain-town internet service. The relay would hold line of sight over broken terrain; the cabinet below would aggregate homes and shops; a fibre would carry their traffic toward Latacunga or another upstream handoff. If the radio lost power, if fog or alignment damaged the link margin, or if the cabinet’s feeder fibre were cut, one physical break could darken the lights of an entire cluster of customers.

That is a useful failure scenario, but it is not a documented description of CORPORACIÓN LORIATELECOM S.A.S. The most consequential public engineering statement points in the opposite direction. ARCOTEL’s September 2024 grant, Resolution ARCOTEL-CTHB-CTDS-2024-0188, says that the company’s internet-access application did not include wireless networks. The appendix does not disclose a radio count, frequencies, tower coordinates, line-of-sight paths, antenna gains, backup power or a radio-to-fibre handoff. It also does not publish the physical network design behind the service. The correct conclusion is therefore narrow: the grant supports a corporate internet-access authorisation, while a mountain relay chain remains unproven.

The current commercial evidence is explicitly fibre-led. The company’s June 2025 tariff sheet describes prepaid monthly fibre to the home, or FTTH, with symmetric download and upload rates. It does not advertise fixed wireless as the delivery medium for those plans. This matters because a fibre access network has a different concentration pattern from a radio chain. Risk moves into feeder fibres, pole routes, splice closures, passive splitters, optical line terminals, aggregation switches and the power feeding the active equipment. There may still be radio links elsewhere in the operating environment, but the public record reviewed here neither identifies nor locates them.

The company’s online coverage layer is similarly noncommittal. Its public KML file contains a single placemark called “Zona de Cobertura” and a polygon. It contains no tower, cabinet, office, OLT, fibre segment, handhole, pole, splice point, route name or capacity attribute. A shaded area can tell a prospective customer where an installation may be worth asking about. It cannot show whether two towns share one feeder, whether a hilltop exists in the chain, or how traffic would be rerouted after a break.

The opening image, then, should be read as a disciplined question. Where does radio, if any, end? Where does fibre begin? Which active site needs power? How many customers sit behind the handoff? What spare hardware can a field technician reach after dark? For LoriaTelecom, the public answer begins with FTTH and ends before the topology starts.

The operator, brand and contract do not sit on one date

The strongest current identity evidence links the legal operator and the retail brand. The company’s May 2025 privacy policy names Corporación Loria Telecom S.A.S., gives Ecuadorian tax identifier 0591762997001 and says it serves customers under the Fiber Spot brand. The same document describes customer installation addresses, billing, support and technical analysis, and contemplates contracted parties participating in installation, support or collection work. Those statements support a present operating relationship between the corporation and the brand. They do not disclose which network assets the corporation owns, which it leases, or how much field work is outsourced.

The public model contract is older and materially different. The contract document still linked from the company site identifies Carla Estefanía López Chango as an individual provider trading as Loria Telecom, with tax identifier 0503032559001. The website’s documents page dates that file to 2020, whereas the corporate privacy and tariff documents were updated in 2025. This is not evidence that the current service is unauthorised or that customers necessarily receive the old form. It is evidence that a prospective customer cannot use the public contract alone to determine which legal person signs today, which tax identifier appears on the bill, or whether the contractual service levels have been conformed to the 2024 corporate title.

The resource registry supports the corporate side of the boundary. LACNIC’s AS273035 registration is active and lists the corporate organisation as registrant. The corresponding organisation record names CORPORACIÓN LORIATELECOM S.A.S. and a Saquisilí address. The associated technical and abuse contact record names Miguel Soria and uses a fiberspot.com.ec address. These records connect the legal company, internet number resources and the Fiber Spot domain. They are strong evidence of control over the registered network identity, not a title register for poles, ducts, offices, OLTs or cabinets.

The date boundary matters. ARCOTEL granted the corporation a 15-year internet-access title in September 2024. The old contract predates that grant. The tariff, privacy and quality files postdate it. A careful account should therefore distinguish at least three layers: the earlier Loria Telecom trading activity associated with an individual; the current corporate operator named in regulator and registry records; and the Fiber Spot customer-facing brand. They appear related and continuous, but continuity of branding does not automatically settle succession of every asset, customer agreement or liability.

For households, municipalities and suppliers, this is not clerical trivia. The contracting entity determines who owes the service, who receives a formal complaint, who must maintain the network and who bears the obligation to restore it. The operator boundary is reasonably well supported for 2024-25; the contract boundary remains stale in the public materials. A current signed-form contract, current billing identity and explicit asset-operation statement would close that gap without requiring disclosure of commercially sensitive network details.

Four storefronts anchor a service area, not a topology

Fiber Spot’s current home page names four public-facing locations: Saquisilí at Simón Bolívar and Mariscal Sucre; Mulaló at 27 de Febrero; Pujilí at Vicente Rocafuerte and Belisario Quevedo; and Latacunga at Avenida Rumiñahui and Quijano y Ordóñez. It also markets residential internet, corporate internet, structured cabling, cameras and alarms, and says support is available around the clock. These are meaningful signs of an active local service organisation. Four addresses create places where a customer can seek sales or support, and their distribution is consistent with a Cotopaxi-focused business.

They do not, however, identify four network nodes. An office can contain an OLT and aggregation router, or it can be only a counter with customer-service staff. A node can sit in an unmarked equipment room, a cabinet, a leased facility or another operator’s point of presence. The company’s about page states a mission and vision but does not distinguish commercial offices from technical sites. No public inventory says which location has active network equipment, generator or battery backup, route diversity, spare optics, switching capacity or a staffed repair depot.

The implementation of the coverage map reinforces that distinction. Fiber Spot’s map script centres a hybrid map near latitude -0.859666 and longitude -78.634955 at zoom level 11, then loads the one KML layer. It does not load a second layer for access nodes, backbone routes or service offices. The latest visible GitHub commit to the KML, dated February 2025, says the file was adjusted for better centring, zoom and a coverage label. That makes the map a recent customer-facing artefact, but the edit history describes presentation rather than a surveyed infrastructure update.

Geography still shapes the likely operating burden. The polygon runs roughly 40 kilometres north to south and about 23 kilometres east to west at its widest coordinate span. It embraces an elongated highland service area rather than one dense urban grid. That shape can imply longer travel times, more exposed pole plant and multiple local concentration points, but those are analytical possibilities, not mapped facts. Exact road distance, elevation, fibre path and crew travel time cannot be derived from the polygon.

There is also a legal-versus-operational distinction. The 2024 title authorises internet access at national scope; the named offices and public coverage layer point to a much narrower present commercial footprint in Cotopaxi. A national authorisation permits expansion. It does not prove that the network currently reaches the rest of Ecuador. Conversely, the polygon does not guarantee service at every address inside it: the tariff sheet says installation remains subject to technical feasibility.

The most defensible service-area statement is therefore modest. Fiber Spot publicly solicits and supports customers around Saquisilí, Mulaló, Pujilí and Latacunga, and it displays a broad connecting coverage area. The physical network tying those places together—if it is one network rather than several branches sharing upstream services—remains unpublished.

One polygon covers a valley but maps no plant

The company’s documents page is unusually useful for dates and customer obligations: it links the contract, tariffs, service-quality policy and privacy policy. Yet it offers no network diagram or infrastructure report. The public coverage map fills that visual space with a single area, not an asset map. This distinction should govern every inference made from it.

The KML polygon has hundreds of coordinate points, which can create a false impression of precision. Those points precisely trace the border of the drawn shape; they do not precisely locate serviceable premises or physical plant. There are no line strings that could be interpreted as fibre routes and no point features that could be interpreted as nodes. There is no metadata stating when a street was passed, whether an address is connected, whether a section is aerial or buried, or whether the border reflects radio propagation, sales policy, administrative limits or hand drawing.

A local report supplies context, but not current topology. In August 2022, Al Día Online reported that Fiber Spot had opened a fourth office or “node” in Latacunga after Saquisilí, Pujilí and Mulaló, served more than 3,000 customers and used fibre carried on poles. The article names people associated with the business and describes local expansion. It is valuable historical evidence of brand activity and an aerial-fibre model. It predates the September 2024 corporate title, and its wording does not establish whether each office was an independently powered aggregation node, whether the reported customer count referred to active paying lines, or whether the same count remains relevant four years later.

The word “node” is especially easy to overread. In local commercial reporting it may mean an office, a branch, a service point or a technical aggregation location. Without equipment lists, rack photographs, power records or route data, it cannot carry all those meanings at once. The Latacunga location could be a major upstream handoff; it could also be a sales branch connected back to another town. The same uncertainty applies to the other three locations.

No exact inter-town route should be drawn from these materials. One cannot responsibly trace a fibre from Mulaló to Latacunga, place a cabinet beneath a hill, or identify a shared bridge, road or pole corridor. Nor can the broad polygon establish route independence. Two paths that appear to approach a town from different directions on a conceptual drawing may converge on the same utility poles, duct, bridge crossing or upstream building. A genuine resilience map would identify at least the endpoints, route classes, shared-risk segments and failover direction while allowing sensitive details to remain generalised.

The map is therefore strong evidence of intended market coverage and weak evidence of engineering. It helps answer “where might Fiber Spot sell?” It does not answer “what breaks together?” In a mountain-town network, that second question is the one that determines whether a single landslip, vehicle strike, transformer outage or cut feeder can isolate an entire valley.

800 Mbps is a retail ceiling, not aggregate capacity

The tariff sheet gives unusually clear retail numbers. Fiber Spot’s four residential plans are symmetric: 200/200 Mbps for US$20 a month, 400/400 Mbps for US$25, 600/600 Mbps for US$35 and 800/800 Mbps for US$45, including tax. Service is prepaid monthly. Installation is advertised as free only within 300 metres of the nearest Fiber Spot network point and remains subject to geographic and technical availability. These terms tell a customer what access rate can be provisioned at an eligible home. They reveal almost nothing about capacity above that home.

An 800 Mbps optical access profile is not proof that 800 Mbps is continuously available through every layer. The usable result depends on the customer router and Wi-Fi, the optical network unit, the passive split, the OLT port, local aggregation, upstream handoffs and internet paths. A provider can sell many access ports whose headline rates exceed the shared backhaul because customers do not all transmit at maximum speed at once. That is normal statistical multiplexing. The important questions are the contention assumptions, peak utilisation and the margin retained for failure conditions.

The company’s service-quality policy sets targets including an average effective fault-repair time of no more than 24 hours and resolution of 98% of general complaints within seven calendar days. It also describes complaints concerning bandwidth below 98% of the contracted rate and says quarterly reports will be supplied to ARCOTEL. These are commitments and measurement rules, not published results. No public quarterly table reviewed for this article shows achieved speed, latency, packet loss, outage minutes, complaint volume or repair performance by town.

Operational activity is corroborated from another angle by a March 2025 Universidad Técnica de Cotopaxi project. The repository record describes a billing and payment application for Fiber Spot by Loria Telecom in Saquisilí, while the full academic report discusses an interface supplied by the company and a customer payment-confirmation process. This supports the existence of a functioning subscriber and billing operation in 2025. It does not disclose active line count, busy-hour traffic or network headroom.

Installed, lit, sold and usable capacity must therefore remain separate. The advertised plan is sold access capacity. The registered prefixes are address resources. The BGP routes show reachability. None is a measurement of the upstream ports or the optical access system. There is no public OLT count, PON generation, split ratio, feeder-fibre count, aggregation-switch port inventory, transit commitment, burst allowance, 95th-percentile use or emergency reserve.

The absence becomes more important during a failure. Suppose two upstreams each carry normal traffic but one link fails. If the surviving port has enough spare capacity, customers may see little more than a routing change. If it is already heavily loaded, logical failover can produce severe congestion. Likewise, a spare OLT card is useful only if it is compatible, configured and reachable; unused fibre is useful only if it follows a different risk path and can be lit. Fiber Spot’s prices and access rates establish a competitive retail proposition. They do not establish the safety margin behind it.

AS273035 proves a live edge, not a handoff location

The public routing evidence is stronger than the physical map. LACNIC assigns the company the active IPv4 block 168.232.100.0/23, covering 512 distinct IPv4 addresses, and the IPv6 block 2803:2b90::/32. The IPv4 record associates the block with AS273035 and the corporate registrant. These are real number resources under the company’s registered control. They do not indicate how many addresses are assigned to customers, infrastructure, carrier-grade NAT systems or internal services.

RIPEstat’s routing-status view, queried for 16 July 2026, reports that the autonomous system was first seen in its observation set on 13 March 2024 and remained visible on 16 July 2026. It counted three IPv4 announcements covering 512 addresses and two IPv6 announcements covering the equivalent of 65,536 /48 networks. The latter is a way of expressing the size of a /32 allocation, not a count of deployed customer networks.

The announced-prefix history showed five current entries during the 2-16 July 2026 observation window: the covering IPv4 /23, its two /24 more-specifics, the IPv6 /32 and an IPv6 /40 more-specific. The overlapping IPv4 announcements must not be added together. The /24s sit inside the /23, so the unique IPv4 space remains 512 addresses rather than 1,024. Similarly, the IPv6 /40 sits inside the /32 and does not enlarge the allocation.

This pattern can be operationally useful. More-specific announcements may steer parts of the address space toward different upstreams or provide finer control during faults. But route visibility does not locate the routers. The BGP session could terminate in Latacunga, Quito, a carrier facility, a leased rack or a remote service delivered over another transport. No public record reviewed here gives a handoff address, interface capacity, cross-connect identifier, facility name or last-mile carrier for either adjacency.

Nor does the autonomous system prove that every Fiber Spot customer uses these prefixes. Residential subscribers may sit behind private addressing and carrier-grade translation; corporate customers may use other arrangements; some services may be delivered through upstream-assigned space. Conversely, an announced prefix can remain visible while parts of the access network are offline. Internet collectors see the route to the operator’s edge, not the condition of a splitter, cabinet or drop cable in Pujilí.

The correct status conclusion is still important: this is not merely a brand website with no visible network identity. AS273035 is active, its company-held IPv4 and IPv6 resources are being originated, and the routes are broadly visible. That supports current network operation at the internet edge. It cannot settle the physical path from a customer optical terminal to that edge, the number of aggregation domains behind it or the capacity available at the handoff.

Two upstream ASNs improve the logic, not necessarily the physics

RIPEstat’s ASN-neighbour view identified three left-side neighbours on 16 July 2026: AS263238, UFINET’s AS52468 and private AS65001. The private ASN appeared with only one observing peer and is not a globally routable public network identity, so it should not be treated as a third independent transit provider. The two meaningful public adjacencies are AS263238 and AS52468.

The BGP-state observations make the division visible. Paths to 168.232.100.0/24 commonly ended through UFINET AS52468, while paths to 168.232.101.0/24 and the covering /23 commonly ended through AS263238. IPv6 paths were visible through both public neighbours. This is materially better evidence than a generic claim of “redundancy”: two distinct upstream ASNs are observable in global routing, and different prefixes are not confined to one public neighbour.

Independent network directories broadly corroborate the pair. The IPIP AS273035 record lists AS52468 and AS263238 as upstreams, while the IP2Location network page also presents two upstream relationships and no downstream network. Those services are secondary summaries and can lag or simplify routing behaviour, but their agreement with live collector data increases confidence that the adjacencies are not a transient display error.

The two upstreams also have different public identities. PeeringDB’s AS52468 page describes UFINET as a regional network service provider. The AS263238 relationship page shows AS273035 as a downstream network. Neither page identifies LoriaTelecom’s local circuit, port speed, contractual protection, handoff building or fibre provider. The scale or reach of an upstream’s wider network must not be transferred to one customer link without evidence.

This is where logical diversity and physical diversity diverge. Two BGP sessions can travel over two genuinely independent carriers, enter different facilities, use separate power systems and survive a local cut. They can also share the same roadside fibre, bridge, duct, pole line, carrier metro ring, building entrance, rack power or upstream transport before separating logically. Even if one /24 is normally preferred through UFINET and another through AS263238, the two paths may still fail together in Cotopaxi.

The public route data does not show whether both upstreams can carry the full customer load. It does not show whether failover is automatic, whether default routes and return paths converge cleanly, whether the surviving carrier accepts all prefixes, or whether the operator has tested a complete withdrawal. Repeated origin prepending on some observed paths suggests traffic-engineering intent, but it is not a disaster-recovery report.

The right claim is therefore “two meaningful public upstream adjacencies,” not “two independent physical routes.” That is a significant piece of resilience architecture, but only one layer. Physical entry diversity, separate transport ownership, handoff geography, power independence and tested spare capacity are still required before a customer can infer that one cut will leave the service usable.

Route hygiene is mixed while installed capacity stays unknown

Routing quality is not binary. RIPEstat’s routing-consistency view found the covering IPv4 /23 and IPv6 /32 both in BGP and route registries, while the two IPv4 /24s and IPv6 /40 were visible in BGP but not in the corresponding route-record view. It also did not find documented route-policy entries for the observed peers. This does not make the live routes illegitimate. It means the public registration picture is less complete than the announcements themselves.

Resource Public Key Infrastructure provides another split result. The RPKI check for 168.232.100.0/23 was valid for origin AS273035 with maximum length /24. That authorisation covers the /23 and its two /24 more-specifics, reducing the risk that networks enforcing route-origin validation will reject those IPv4 announcements as invalid. By contrast, the RPKI check for 2803:2b90::/32 returned “unknown” because no validating route-origin authorisation was found. Unknown is not invalid, but it lacks the positive cryptographic origin statement present for IPv4.

The company also had no returned network record in a current PeeringDB ASN query. That absence means there is no self-published PeeringDB profile exposing traffic levels, facilities, exchange points, policy or public interconnection contacts through that interface. It does not mean the company lacks peering or transit. Small regional operators often rely on private carrier arrangements and do not maintain a public profile.

A CIDR Report view likewise identifies the two public upstream adjacencies. Its aggregate address-space figure should be handled cautiously because counting overlapping covering and more-specific routes can exaggerate unique IPv4 space. Registry boundaries, not a sum of every visible prefix line, establish the distinct allocation.

None of these route-hygiene facts reveals installed bandwidth. A valid origin authorisation proves who is allowed to originate a prefix, not whether the associated port is 1, 10 or 100 Gbps. A registered route object helps other networks construct filters, not size a backhaul. Multiple paths prove reachability choices, not headroom. The distinction is particularly important for an ISP selling 800 Mbps plans: a modest upstream can support many such customers under normal usage but struggle badly when one link fails or evening demand spikes.

The missing capacity schedule would separate at least six states: designed capacity, physically installed ports and optics, lit capacity, contractually committed capacity, traffic already sold to customers and capacity usable after the largest single failure. Public evidence supplies none of those numbers. It also lacks the number of OLT ports, active PON branches, split ratios and maximum customers per branch. The route edge is demonstrably live and reasonably organised; the load it can safely carry remains unknown.

Poles, cabinets, power and labour define the outage clock

For customers, most outages begin below the BGP layer. The 2022 local report’s description of fibre on poles is plausible for a network spanning Cotopaxi towns, and the tariff’s 300-metre installation condition implies a distributed set of network-near points. But the current public record does not inventory pole agreements, feeder routes, splice closures, cabinets, optical splitters or the ownership boundary between the provider and utilities. It also does not say whether critical feeder sections are buried, aerial or duplicated.

This leaves several physical failure modes open. A vehicle can break a pole carrying multiple fibres. Wind, falling branches or construction can damage an aerial span. A single closure can collect several branches whose customers appear geographically dispersed. An aggregation cabinet can remain physically intact but fail when its power supply, battery, switch or environmental control fails. A route can be repaired quickly if spare cable, closures, optics and trained staff are nearby, or remain down while materials and access permissions are assembled.

The company’s public promise of 24/7 support is useful, but a telephone response is not the same as a field response. The quality policy’s 24-hour average effective fault-repair target also excludes some external causes and describes an average rather than a guaranteed restoration time for every incident. No published staffing plan identifies the number of fibre splicers, climbing-qualified technicians, vehicles, shifts, contractors or depots available across the four named towns.

The privacy policy’s reference to contracted installation and support parties suggests that some work may be performed through partners, but it does not state availability or service-level terms.

Three public local-government purchase registers add a narrow operating signal. Mulaló’s September 2023 register, November 2023 register and February 2024 register each contain a small payment line associated with the corporation’s tax identifier. These records support transactions between the company and the parish government around that period. They do not specify the service purchased, prove a network node at the government site, or reveal the size of the local customer base.

Power is the largest unmeasured common dependency. Passive fibre splitters need no electricity, but OLTs, aggregation switches, edge routers, cooling and any radio equipment do. The public materials do not identify the utility feeds for technical sites, battery autonomy, generator capacity, refuelling plan, remote monitoring or low-voltage shutdown behaviour. An office address does not prove that a technical node has independent backup; a tower image would not prove that its cabinet does.

Recovery evidence should therefore be operational, not decorative. It would state which classes of sites have batteries, the minimum tested runtime under current load, whether generators cover core locations, how alarms reach on-call staff, where compatible spare optics and radios are held, and the target travel time to each service cluster. If there are no radios in the service chain, spare-radio questions should be closed explicitly rather than answered with assumptions. Until then, the outage clock is bounded by unknown power autonomy, unknown shared pole routes and unknown field capacity.

Customer scale is the missing denominator

Network concentration cannot be evaluated without a denominator. The 2022 article’s “more than 3,000” customer claim provides a historical scale marker, but it is too old and too loosely defined to serve as a current subscriber count. It predates the corporate title, the 2025 tariffs and the current route picture. It does not say how many lines were active, how many were residential or corporate, whether customers were counted by account or service, or how they were divided among the four towns.

A commercial company profile offers another signal. EMIS describes Corporación Loriatelecom S.A.S. as a Saquisilí-based wired-telecommunications company incorporated in July 2022 and reports a small employee count and substantial recent percentage changes in revenue and assets. Because the detailed financial statements are behind a commercial service and the percentages can rise from a small base, the figures should be treated as third-party indicators of formal activity, not audited proof of network scale or investment capacity.

Internet measurement can also be misread as customer data. An APNIC economy-level measurement page includes AS273035 in estimates derived from observed samples. Such estimates can suggest that traffic from the autonomous system reaches end users, but samples, address observations and market-share estimates are not subscriber records. They cannot determine how many Fiber Spot bills exist or how many customers share one feeder.

The same warning applies to route-enrichment services. A BigDataCloud lookup for an address inside the company’s IPv6 space surfaces the public upstream context and can echo the private-AS artefact seen by collectors. It is useful corroboration of network identity, not an inventory of deployed IPv6 customers, access equipment or physical links.

What matters for resilience is not merely the total customer count but its distribution. If 1,000 customers share one OLT uplink, one feeder or one unprotected cabinet, the impact of a single failure is large. If the same customers are divided across several independently powered sites and diverse routes, the blast radius is smaller. A total without the number of handoffs, branches and customers per branch cannot reveal that difference.

The missing denominator also affects economics. Plans priced from US$20 to US$45 create recurring revenue, but the ability to fund route diversity, backup power, spares and round-the-clock field coverage depends on take-up, churn, collection, customer density and the cost of upstream capacity. A broad rural or peri-urban footprint can be socially valuable while making duplicated routes expensive. That trade-off should be analysed with current counts, not guessed from a polygon.

A useful public disclosure would not need customer names or exact household coordinates. Town-level active line ranges, customers per aggregation site, peak concurrent load, branch utilisation bands and the largest single failure domain would be enough to transform the resilience discussion. Without them, the operator appears active and locally established, but the number of people behind each handoff remains unknown.

The evidence needed to prove a recoverable network

CORPORACIÓN LORIATELECOM S.A.S. has already cleared several important evidentiary thresholds. Current corporate documents identify Fiber Spot as its brand. ARCOTEL granted a long-duration internet-access title. LACNIC records associate the corporation with AS273035 and active IPv4 and IPv6 resources. Public collectors see those resources through two meaningful upstream networks. A current plan sheet sells symmetric FTTH in four tiers, and the company names four Cotopaxi offices. These facts support a live regional ISP, not a dormant shell or an unsupported directory label.

The unresolved question is whether the network can keep serving customers through a material local failure. The quickest way to answer it would be a bounded engineering disclosure rather than a promotional claim. First, identify the access architecture by service zone: FTTH only, or FTTH plus any radio backhaul. If radios exist, state the number of operational relay hops, general service zones, licensed or unlicensed status, line-of-sight design margin, power autonomy and spare-radio policy. If none exist, say so and retire the relay hypothesis.

Second, publish a generalised physical topology. It need not expose exact security-sensitive coordinates. It should distinguish offices from technical sites; mark primary aggregation zones; show whether Saquisilí, Mulaló, Pujilí and Latacunga are connected in a ring, a chain or separate branches; and identify shared-risk corridors. The single current polygon cannot perform that function.

Third, disclose capacity by state. For each upstream, give the installed port rate, committed rate, normal busy-hour peak and capacity available after the other upstream fails. For access, give OLT and PON generation, branch counts, maximum split, current utilisation bands and the policy for adding capacity. The exact customer traffic mix can remain confidential; headroom and failure-condition usability are the relevant facts.

Fourth, define the recovery envelope. List site classes with tested battery runtime, generator coverage, alarm monitoring and refuelling arrangements. State the number and location range of field teams, whether splicing is in-house or contracted, standard spare holdings and escalation targets. The public LACNIC contact role establishes a technical and abuse point of contact for the internet resources, but it is not a field-restoration plan.

Fifth, reconcile customer-facing documents. Replace or clearly archive the 2020 individual-provider contract, publish the form used by the corporation, and release the quarterly quality results contemplated by the 2025 policy. Report achieved repair time and speed compliance by service zone, with external-cause exclusions shown separately. That would allow readers to distinguish an objective from actual performance.

Finally, quantify concentration. Current active lines by town, customers per aggregation point, the largest feeder failure domain and the number of customers that remain serviceable after the loss of one upstream or one core site would answer the assignment’s central question. Those numbers would also reveal whether local support labour and spare inventory are proportionate to the network’s real exposure.

Until such evidence appears, the balanced judgement is neither that Fiber Spot lacks redundancy nor that it has proved resilience. The internet edge has credible logical diversity. The retail FTTH offer is clear. The physical routes, facility roles, power systems, access loading and recovery resources are not. The imagined sunset relay remains a useful symbol precisely because the record cannot say whether it exists. What customers can see is the lit valley and the sales promise; what they still cannot see is the chain of assets, people and spare capacity that keeps the connection alive after one link goes dark.