Summary
- Metrowireless’s public materials establish a real Costa Rican fixed-internet operator using point-to-point and point-to-multipoint wireless links, with fibre also offered, but they do not disclose tower coordinates, radio-hop diagrams, fibre routes, upstream commitments, site power systems or installed aggregate capacity.
- Its published coverage is best read as a feasibility envelope, not proof of service at every address: a fixed-wireless connection depends on line of sight, usable shared spectrum, powered equipment at both ends and a surviving handoff beyond the radio access network.
- The strongest resilience test is operational rather than promotional: verify each relay and fibre handoff, measure busy-hour headroom and failover, document who can enter each roof or pole site, and time restoration from customer loss of service through radio, power, transport and upstream escalation.
The first alignment is the first dependency
The Santa Ana roof is not a documented Metrowireless installation, and the technician at first light is a representative scene rather than an account of a particular customer. It is nevertheless the right place to begin because the company describes the physical logic itself. Its wireless-home explainer says that a microwave signal travels from a tower to an antenna outside the customer’s property, that line of sight is needed, and that a technician must identify a suitable installation point. A cable then enters the building and reaches powered indoor equipment. The radio link is not an abstract last mile. It is a chain of sight, mounting, cabling and electricity.
That chain changes the meaning of “availability”. A fibre drop can be constrained by ducts, poles and civil works; a fixed-wireless link can cross ground quickly, but only if the path clears roofs, vegetation and terrain. The International Telecommunication Union’s terrestrial line-of-sight design recommendation treats clearance, diffraction, multipath, rain attenuation, outage probability and diversity as engineering variables. A view that looks open to a person is not necessarily enough: the radio path has a volume around its centre line, and partial obstruction can degrade it even when the antenna faces the correct direction.
The company’s main site says it transports data and internet over wireless or fibre and operates from Santa Ana. That statement establishes the business model’s two media, not the topology joining them. The rooftop antenna could point directly to a high aggregation site or to a relay that forwards traffic through another radio hop. That site could hand off to fibre locally or carry traffic farther before it reaches a fibre connection. None of those alternatives can be chosen from public evidence with confidence.
This distinction matters because each alternative creates a different failure domain. A direct customer-to-aggregation link can isolate one subscriber when the customer antenna moves. A relay serving several sectors can turn one power failure, damaged mount or blocked path into a neighbourhood outage. A shared fibre handoff can concentrate many otherwise healthy radio links behind one cut or equipment failure. Even the neatest customer installation says nothing by itself about diversity beyond the first visible endpoint.
The same is true of recovery. A customer can restart indoor equipment, but cannot realign a rooftop radio safely. A field technician may reach the customer roof but lack immediate access to the far-end property. The radio team may restore the link while the aggregation site remains without power. The access network may be intact while the fibre carrier or internet upstream is still down. The first alignment is therefore both a service-enabling act and the first allocation of responsibility across a chain the customer cannot see.
Metrowireless’s local value is clearest here. A regional operator can know the roofs, ridges, landlords and travel times of its territory better than a remote support desk. Yet local knowledge is not equivalent to physical redundancy. It can shorten diagnosis and dispatch; it cannot create a second path, spare power or unused radio capacity unless those resources were installed beforehand. The public record supports the presence of technicians and a fixed-wireless operating model. It does not reveal how much redundancy stands behind either one.
What the public record actually establishes
Metrowireless is more than a trading name on a website. Costa Rica’s telecommunications regulator maintains a current operator listing for Metrowireless Solutions de Costa Rica, S.A., associated with legal identifier 3-101-589655 and authorization RCS-003-2023. The regulator’s contract registry page identifies a current adhesion contract for fixed internet and links. Its approved contract cover describes fixed internet through point-to-point and point-to-multipoint wireless networks, leased lines and virtual private networks for residential, small-business and corporate customers.
The final regulatory action is unusually useful because it fixes the boundary between a company’s claims and its legal permission. In RCS-003-2023, the regulator granted authorization for public telecommunications services including fixed internet over point-to-point and point-to-multipoint networks, leased lines and virtual private networks. The legal instrument supports the statement that these are authorized activities. It does not certify that every authorized service is present in every named district, that every tower shown by marketing is live, or that a customer can be installed without a site survey.
An earlier official market review of wholesale dedicated lines also placed Metrowireless in a market that included dedicated internet, local-area-network interconnection, backup and related business services. That is historical evidence of the company’s role and service orientation. It should not be converted into a 2026 inventory. Product portfolios, customers, suppliers and installed equipment can all change after a regulatory market study.
The legal and commercial boundaries are also not identical. The customer signs with Metrowireless, but the service may depend on property owners, pole infrastructure, electrical utilities and external network providers. A regulator can authorize the operator and approve customer terms without owning those dependencies or guaranteeing their performance. Conversely, a supplier or property owner can affect restoration even though it has no retail relationship with the affected customer.
There is another important boundary in the spectrum record. A registry entry for a direct-use concession associated with resolution 176-2020 records that concession as extinguished through a 2025 executive action. The operator list still exposes a reference to the older concession alongside the later service authorization. Those records should not be blended into a claim that the company currently holds a particular exclusive radio channel. They concern different legal instruments, and the public service authorization itself refers to networks operating in free-use spectrum.
The defensible entity description is consequently narrow. Metrowireless Solutions de Costa Rica, S.A. is an authorized Costa Rican provider of fixed internet and data connectivity, with regulatory approval for point-to-point and point-to-multipoint wireless delivery and related links. Public materials also advertise fibre. The evidence does not establish ownership of every support structure, a proprietary nationwide fibre network, an exclusive spectrum position, or control over all facilities between a customer and an international destination.
That narrow description is more useful than a broad one because it shows where resilience questions belong. Metrowireless can be examined for the design, maintenance and restoration of the equipment and services under its control. Claims about a pole owner, electricity distributor, fibre wholesaler or upstream network require separate evidence. An outage may cross all of those boundaries, and a credible account must say where observation ends rather than assigning every dependency to the retail operator.
The authorised footprint is wider than the proven physical network
The broadest geographic evidence is an official 2022 authorization notice in La Gaceta. It names extensive service areas across cantons in San José, Alajuela, Cartago and Heredia, plus parts of Puntarenas, for wireless networks using free-use bands. It separately describes a much narrower wired-network scope in Guácima and three districts of San Ramón. The notice is a legal permission map. It is not a list of commissioned towers, connected buildings or lit fibre kilometres.
Metrowireless provides a more granular public artifact: an interactive fixed-internet coverage map. The page says the blue areas indicate wireless presence, but it also states that the map is predictive and that actual availability is subject to technical feasibility. That caveat is central. A prediction can guide sales qualification; it cannot see a new building, a mature tree, a landlord’s refusal, a weak mounting position or local interference at a specific roof.
The map loads a public June 2024 KML file. A direct inspection on 17 July 2026 found 97 polygon placemarks and 3,130 coordinate pairs, with an overall coordinate envelope extending approximately from longitude -84.700937 to -83.662707 and latitude 9.637279 to 10.213815. Those counts describe the file, not the current network. The placemarks carry generic polygon names. They do not identify towers, antenna heights, azimuths, frequencies, radio capacities, fibre handoffs, relay order, owners or commissioning dates.
The KML therefore answers one question well: where did the operator publish modeled areas of potential wireless service in that file? It cannot answer the more consequential questions: how many radio sites were operating, which polygons were served by the same site, where traffic converged, whether the underlying equipment remained live after June 2024, or whether any path had a diverse alternative.
The legal notice and KML also describe different things and should not be forced to match. Authorization can cover districts where no equipment is deployed. A modeled polygon can occupy only part of an authorized canton. A live installation can still fail feasibility inside a modeled area. Conversely, a later network extension might not appear in a dated file. The gap between legal scope, published prediction and installed plant is not evidence of wrongdoing; it is a normal distinction that matters whenever geographic marketing is used to infer resilience.
For a physical map fit for operational analysis, at least five additional layers are needed. The first is verified site location, with tower or rooftop identity and access conditions. The second is radio geometry: sector orientation, customer or relay role, and whether a hop is a single point of failure. The third is transport: fibre handoff, route and provider. The fourth is power: utility feed, battery or generator support and tested runtime. The fifth is demand: customers and busy-hour traffic dependent on each site. None is recoverable from blue polygons alone.
This absence limits any estimate of affected users. A large polygon may contain few subscribers; a small urban sector may carry many. One relay could serve several polygons, or several sectors could overlap within one. The public material gives no customer count by district, no subscriber concentration by tower, and no separation between residential, small-business and corporate endpoints. It is possible to describe the exposed classes of users, but not to quantify them responsibly.
The map is nevertheless valuable because it prevents two opposite errors. It rejects the idea that Metrowireless has no visible geographic claim at all, while its feasibility warning rejects the idea that colored territory equals installed availability. Resilience work should start from that middle ground: a substantial modeled footprint exists, but the physical graph underneath it remains unresolved.
The radio path: clear air, shared spectrum and rain
A fixed-wireless hop has no trench to cut, but its transmission medium is not empty. Buildings and vegetation can enter the path; mounts can move; antennas can lose alignment; other transmitters can raise the noise floor; rain can add attenuation at frequencies where precipitation matters. These mechanisms differ in speed and diagnosis. A new obstruction can create persistent degradation, interference can fluctuate with other users, and a severe storm can combine propagation loss with power or structural damage.
Costa Rica’s Telecommunications Law treats free-use spectrum differently from exclusive concessions. Article 9 of Law 8642 says specified bands may be used without a concession, authorization or permit, subject to the applicable technical framework. The national frequency plan’s free-use provisions set conditions and identify multiple eligible ranges. Neither record identifies which band, channel width, polarization or transmit setting Metrowireless uses on any particular hop.
The operating trade-off appears in the implementing rules. Costa Rica’s regulation for free-use equipment makes clear that such systems must tolerate harmful interference from other authorized operation and do not receive protected status in the manner of an exclusive assignment. That does not mean interference is present on a Metrowireless link. It means that a resilience assessment cannot assume exclusivity merely because the access network is professionally installed.
Interference resilience is designed through site selection, channel planning, antenna directivity, link margin, monitoring and the ability to change configuration or path. Public evidence supplies none of Metrowireless’s per-hop values. There is no disclosed spectrum scan, fade margin, modulation floor, channel reuse plan, alarm threshold or spare-channel policy. As a result, it is possible to identify interference as a mechanism, but not to calculate the probability that it will disrupt a particular customer.
Rain requires the same discipline. The company says its residential wireless link is calibrated for Costa Rican rain, which is a useful description of intended design. The ITU’s rain-attenuation model explains how specific attenuation depends on rainfall rate, frequency and polarization. Without the band, path length, fade margin and local design target, “calibrated” cannot be translated into minutes of expected annual outage or a guaranteed storm threshold.
Weather can also attack the infrastructure without attenuating the radio wave directly. Costa Rica’s meteorological institute documented strong rain and thunderstorms in the Central Valley and Pacific slope, including gusts and the possibility of falling branches and power lines. That bulletin is regional hazard evidence, not evidence that a Metrowireless site failed. It shows why radio, mounting, access and power have to be considered together.
Santa Ana’s terrain adds a local dimension. The national emergency commission’s canton hazard description identifies steep southern sectors and rainfall-related landslide exposure that can affect roads, bridges and lifelines. Again, the document does not locate company equipment. Its relevance is logistical: a clear radio path over difficult ground can continue carrying traffic while road access is impaired, but a failed remote site on that same ground may take longer to reach.
Wireless access is therefore both an avoidance mechanism and an exposure. It can bypass slow or unavailable local civil construction and span terrain that would make a new cable expensive. It also places critical alignment and powered electronics in exposed positions, often on property the operator does not own. The resilience result depends on the exact site and hop, not on the medium’s name.
The roof and tower are power sites before they are network sites
Every radio link in this model has at least two powered ends. Metrowireless’s published customer terms require a 110-volt electrical connection for terminal equipment and recommend an uninterruptible power supply for small-business and enterprise service. That is direct evidence that customer-side electricity is part of service continuity. It is not evidence that all customers install a UPS, maintain its battery or achieve a stated runtime.
The far end is less visible. A tower, mast or rooftop aggregation point needs power for radios, switching and any optical handoff. If the path includes a relay, that relay adds another electrical dependency. Public materials do not state whether Metrowireless sites use dual utility feeds, batteries, generators or remote fuel arrangements. They also do not disclose load, battery age, tested runtime or whether cooling and optical equipment share the same backup.
This missing information prevents a common shortcut: multiplying a published network-availability percentage by the number of hours in a year and treating the result as proven physical autonomy. The terms cite 99.97 per cent availability for the operator’s central network and establish performance and compensation conditions. A contractual target or remedy is important to customers. It does not reveal which devices are included in the “central network”, whether customer power is excluded, or how long a particular radio site can ride through a utility outage.
The distinction between interruption and failure is useful. Costa Rica’s electricity distributor maintains a scheduled suspension service for planned work. A scheduled local outage is not a Metrowireless incident, and it need not affect any company site. It demonstrates that commercial power can be intentionally unavailable for maintenance, sometimes for hours. A resilient operator plans for that ordinary condition as well as for storm damage.
At the customer end, backup can preserve the rooftop antenna and indoor router only if every required device is connected. A powered laptop is irrelevant if the power injector feeding the outdoor radio is on an unprotected outlet. A UPS can also outlast the operator’s far-end battery or vice versa. The service survives only for the overlap of all necessary runtimes, not the longest single runtime in the chain.
At a shared rooftop, recovery authority can be as important as battery capacity. The operator may monitor a loss immediately but still need a keyholder, building manager or safety approval to enter. At a tower, a landlord or maintenance contractor may control access. After a severe event, road conditions can delay travel. A generator does not guarantee continuity if it cannot start, is not fueled, cannot be reached or supplies only part of the load.
The public evidence leaves those questions open. There is no site-by-site power register, no tested-runtime distribution, no stated generator coverage and no published access-time objective. It would be equally unsupported to claim that backup is absent or that it is comprehensive. The correct conclusion is that customer power is explicitly a service dependency, while aggregation-site autonomy remains unverified.
That conclusion changes the most useful measurement. An operator should not be asked merely whether it “has batteries”. The useful evidence is the minimum tested autonomy of every device on a service path, under realistic load and battery condition, paired with the time required to restore utility service or deploy temporary power. For a multi-hop radio path, the weakest powered site governs continuity.
Fibre begins where the radio network stops being wireless
The company describes service over wireless or fibre, but even a wireless customer’s traffic must eventually enter a wired transport and internet-routing environment. A microwave hop can be the access edge, a backhaul segment or both. Somewhere, packets reach switching and optical equipment, then a carrier, exchange or upstream network. That transition is the least documented but potentially most concentrated part of Metrowireless’s physical operating surface.
An official SUTEL record supplies one useful clue. In a 2023 regulatory meeting record, the regulator registered an infrastructure-sharing agreement between Instituto Costarricense de Electricidad and Metrowireless concerning access to and use of support infrastructure, with an effective date in March 2023. The record shows that a formal mechanism for using another organization’s infrastructure existed. It does not identify a pole route, prove that a particular attachment was made, or establish that any customer has a physically diverse alternative.
Pole access is not fibre ownership. A company can attach its own cable, lease fibre, buy a wavelength or purchase an Ethernet service; each arrangement allocates repair and capacity differently. One physical cable can also carry services sold under multiple commercial names. Without route and provider disclosure, the fact that Metrowireless advertises fibre cannot support a claim of an independently owned end-to-end network.
The wired scope in the official authorization notice is geographically narrower than the wireless scope. That difference suggests a business centered on wireless reach with selected wired service, but it still does not reveal where radio traffic meets fibre. The handoff could occur near a tower, at a data facility, at an office or through a third party’s access circuit. Nor does it show whether multiple radio clusters converge on one optical path.
Route diversity has a demanding physical meaning. Two logical circuits are not diverse if they share a pole line, duct, bridge crossing, building entrance, optical distribution frame or carrier core. Two fibre strands in one cable protect against a transceiver failure but not a cable cut. Two providers may still buy the same underlying span. A public route map or letter of diversity would be needed before treating a second service as protection against these common modes.
The handoff also shapes fault isolation. If customer radios remain registered but external destinations disappear, the cause may be an aggregation switch, an optical terminal, a fibre span, the carrier edge or internet routing. Metrowireless can observe portions of that chain, while a supplier may have to test and repair another. The customer sees only loss or degradation. A credible recovery process needs shared timestamps, circuit identifiers, demarcation tests and escalation contacts rather than a generic division between “wireless” and “internet”.
No public evidence identifies Metrowireless’s fibre route kilometres, handoff addresses, optical capacity, suppliers, restoration agreements, spare optics or repair stock. The absence is not evidence of fragility, but it means the fibre side cannot be graded as physically diverse. The correct state is unresolved.
This is where the apparent agility of a radio access network can meet a fixed bottleneck. New customers may be installed without extending fibre to every premises, yet their traffic can accumulate behind the same backhaul. The value of the wireless edge is real; its scale and resilience depend on what happens at the first shared aggregation and every transport segment after it.
BGP shows reachability, not route resilience
Public internet-routing data adds a logical view of Metrowireless, but it must not be mistaken for a trench map. LACNIC’s registration service assigns AS263246 to Metrowireless Solutions de Costa Rica, S.A. It also records an IPv4 allocation covering 190.108.72.0/22 and an IPv6 allocation covering 2800:b90::/32. These are durable resource-registration facts. They do not state how many addresses are in use, how many customers sit behind them, or how much traffic the network can carry.
RIPE NCC’s routing observations provide a current visibility check. Its autonomous-system overview reports AS263246 as announced. The announced-prefixes view observed four IPv4 /24 routes and the IPv6 /32 on 17 July 2026. The routing-status view also showed broad collector visibility and two observed neighboring autonomous systems at the time of the query.
Those observations prove that route collectors could see origins and adjacencies. They do not prove two usable physical exits. One neighbor could be visible only for IPv6, at a limited location or through a relationship that does not carry ordinary customer traffic. Two BGP sessions could cross the same fibre. A backup route could be configured but lack adequate capacity. Collector visibility can also change after the observation time.
A separate CIDR Report view for AS263246 identified the four IPv4 /24s and showed an observed IPv4 adjacency to AS52468, UFINET. One public route probe toward 190.108.72.0/24 also traversed AS52468 before reaching AS263246. The probe is useful corroboration for one measured origin, time and destination. It cannot establish that every prefix, customer or service uses that path, or that no other path exists.
The temptation is to count neighbors and call the result redundancy. That would collapse four layers. The first is registration: who holds the number and addresses. The second is control-plane configuration: what routes are announced and to whom. The third is data-plane behavior: where sampled packets actually travel. The fourth is physical transport: which fibres, buildings, power systems and suppliers those paths share. The public evidence is strongest at the first two layers, offers a narrow sample at the third and is silent at the fourth.
Routing evidence also says little about headroom. An autonomous system can announce every prefix while a congested backhaul drops packets. A failover path can accept routes but carry only a fraction of peak demand. Conversely, a network with one visible transit adjacency may have private transport or peering not exposed by the selected collectors. It would be unsupported to convert the public view into either a claim of single-homing or a claim of full multihoming.
The resilience questions must therefore become more specific. Which external sessions carry IPv4 and IPv6? At which facilities do they terminate? Are the access circuits physically separate? What are the committed and burst rates? Is failover automatic, and has it been tested while the network is busy? Does the backup preserve inbound reachability as well as outbound traffic? Public BGP data can identify where verification should begin, but only configuration, circuit and test records can answer those questions.
For customers, the practical risk is that a network can look healthy from one vantage point while a local cluster is isolated, or look globally visible while its usable throughput collapses. Route visibility is a valuable heartbeat. It is not an inventory of the body that produces it.
Fifty megabits is a sold service, not installed capacity
Metrowireless publishes concrete speed offers. Its residential wireless page lists packages at 30 Mbps downstream and 4 Mbps upstream, and 40 Mbps downstream and 6 Mbps upstream. Its small-business wireless page advertises 50 Mbps downstream with either 10 or 20 Mbps upstream. Its corporate wireless page advertises symmetric 50 Mbps options with different contention descriptions, including a premium one-to-one offer.
Those are retail service parameters, not an inventory of installed network capacity. A 50 Mbps plan can be delivered over a radio capable of more than 50 Mbps, through a sector shared by several customers, into a backhaul shared by several sectors. The customer’s enforced rate, radio’s negotiated rate, sector’s aggregate throughput, backhaul’s line rate and upstream’s committed capacity are five different numbers.
The same distinction applies to “guaranteed”. A one-to-one commercial contention label can describe how a service is sold or dimensioned; it does not by itself establish two physical routes or enough spare capacity after a failure. The company terms state performance conditions including delivery of a percentage of contracted speed. That is evidence of a customer commitment and a basis for remedy. It is not a public measurement of every sector during the busiest hour.
Installed capacity is the sum of equipment and links that exist. Lit capacity is the portion configured and available for use. Usable capacity is lower when protocol overhead, radio conditions, contention, reserved headroom, maintenance state and failure scenarios are considered. Sold capacity is the sum of customer rate commitments, which can exceed instantaneous shared capacity when usage is expected not to peak simultaneously. Resilient capacity is smaller still: it is what remains when the largest credible component fails.
No public record found for this assessment supplies sector channel widths, modulation rates, customer counts per access point, microwave backhaul rates, fibre port speeds, upstream commits or busy-hour utilization. Nor is there a disclosed oversubscription policy that can be applied across residential, small-business and corporate products. The evidence therefore supports the existence and shape of offers, not a total in megabits or gigabits for the company network.
Sector-wide statistics do not fill that gap. SUTEL’s 2024 telecommunications statistics report distinguishes fixed wired and fixed wireless access and describes the national market, but it does not provide Metrowireless’s installed capacity or customer count at the level needed here. The regulator’s 2025 fixed-internet quality report focused its field measurements on four larger providers representing most subscriptions. Metrowireless’s absence from that measurement set is not a negative result; it means that report cannot be used as an independent performance test of this network.
The important denominator is therefore missing. A 50 Mbps customer behind a sector with 500 Mbps of usable busy-hour capacity faces a different risk from the same customer behind a saturated sector. A network with 30 per cent headroom before failure may have none after moving traffic to a backup. Without customer counts, usage percentiles and failure-state tests, even a precise plan speed cannot yield an affected-user or congestion estimate.
A useful capacity disclosure would be modest rather than commercially intrusive. For each aggregation domain, it would show installed and lit access capacity, busy-hour 95th-percentile traffic, the largest single failure, capacity remaining after that failure, and the number of services dependent on the domain. Values could be presented in ranges. What matters is preserving the distinctions between the radio’s headline rate, the customer’s plan and the network’s survivable throughput.
A relay failure concentrates harm unevenly
The likely users of Metrowireless’s network are visible in its approved service categories and current offers: households, small businesses and corporate customers, with historical participation in dedicated connectivity. Their dependency patterns differ. A household may lose remote work, education, communication and entertainment. A shop can lose payment processing, cloud applications or security-camera access. A larger enterprise may have a secondary connection, but can also place more staff and systems behind a single circuit.
Fixed wireless can be especially valuable where wired choices are limited. The residential page presents it as an option beyond conventional cable reach and in mountainous or outlying areas. That value can increase the consequence of failure: customers who chose radio because another fixed route was unavailable may not have an easy substitute. Mobile service can provide temporary access, but its signal, data allowance, indoor reach and congestion characteristics are different and cannot be assumed adequate.
The topology determines how harm concentrates. A moved customer antenna is mostly an individual fault. A failed access sector affects the customers registered to that sector. A relay failure can remove every downstream sector and endpoint that depends on it. An aggregation-switch or fibre-handoff failure can affect multiple radio sites. An upstream problem may be visible across much of the network while local links remain up.
None of the public map artifacts identifies which polygons share a relay or handoff. There is no count of active customers by polygon, sector or service class. Consequently, an estimate such as “thousands affected” or “one neighborhood affected” would be invented. The right public statement is qualitative: concentration could increase sharply at shared relay, power, aggregation, fibre and upstream points, but the magnitude remains unknown.
Local support changes the recovery experience without changing that arithmetic. Metrowireless’s technical-service page describes field visits, appointments and a Costa Rica-based service operation. That supports the existence of local support labor. It does not disclose the number of technicians on duty, their geographic placement, spare-equipment stock, vehicle access, after-hours coverage or the number of simultaneous incidents they can manage.
When several sites fail in one storm or utility event, dispatch becomes a queue. The order may reasonably prioritize a high-concentration relay, critical business service or site whose restoration recovers other paths. A customer waiting at the edge may receive no benefit until a central dependency is repaired. Clear fault-domain information lets the operator choose that order; without it, teams can spend time at customer premises while a shared failure persists upstream.
The published terms say installation, reconnection and repair are generally handled within one business day under stated conditions. That is an accountable service commitment, but it is not the same as a measured restoration distribution during a regional event. One-business-day handling may be excellent for an isolated fault and unattainable when access roads, poles, power and multiple sites are damaged together. Assessment should compare ordinary and severe-event performance rather than force them into one average.
Affected-user analysis should also count indirect dependence. A small wireless circuit can support a point-of-sale terminal used by hundreds of customers, a clinic appointment system, a security service or a local employer’s remote connection. Bandwidth alone is a poor measure of social consequence. The operator may know service class, but only the customer can identify the processes behind the circuit.
A defensible impact register would therefore combine network concentration with customer-declared criticality. It would show endpoints per sector, downstream dependencies per relay, priority services, available alternatives and the restoration authority for each shared site. Until such evidence is available, this article can identify who might be harmed and how, but not how many people a specific failure would affect.
Recovery authority crosses four property lines
Restoration begins with localization. Is the fault inside the customer building, at the rooftop radio, across the air path, at the far-end site, in electrical supply, at an aggregation device, on fibre, or beyond the autonomous system? Each answer moves work to a different person and may cross a different property boundary.
The first boundary is the customer premises. Metrowireless needs permission and safe access to inspect cabling or the outdoor unit; the customer controls indoor power and may control the roof. The second is the radio or relay site, which may be company-operated but located on a landlord’s building or tower. The third is shared support infrastructure, where a pole owner or other infrastructure manager can impose access and safety procedures. The fourth is external transport and upstream service, where another network’s tests and field crew may be necessary.
Those boundaries create a sequence. Remote monitoring should first distinguish loss of customer power from loss of the far end, then check radio registration, signal change, packet loss and the status of shared equipment. If a site is dark, the operator needs utility status and backup-power telemetry. If radios are healthy but destinations fail, it needs demarcation and routing tests. Dispatch should carry the probable spare and the correct access authorization, while supplier escalation proceeds in parallel where responsibility is uncertain.
The sequence is simple on paper but sensitive to records. A technician cannot reach a roof quickly without a current contact and access window. A carrier cannot test the right fibre without a circuit identifier and demarcation location. A replacement radio cannot restore service if its configuration, mount or compatible power unit is unavailable. A generator cannot help if the site load and connection method are unknown. Recovery time is therefore partly a data-quality problem and partly an inventory problem.
The company’s local field-service posture is an advantage when those records are good. Short travel distances and familiarity with property owners can reduce the time between detection and hands-on work. The published one-business-day repair language creates a customer-facing expectation. What remains undisclosed is whether the service organization is dimensioned for correlated failures rather than ordinary appointments.
Recovery also needs an explicit stopping point. Restoring link registration is not enough if throughput remains below a useful level. Reannouncing a prefix is not enough if the backup path is congested. Replacing a customer radio is not enough if an intermittent obstruction or interference condition returns at the same hour. Closure should require stable signal and error measurements, end-to-end traffic tests, confirmation from affected service classes and monitoring through a representative load period.
Post-incident evidence should separate cause, trigger and amplifier. A utility outage might trigger a relay failure; an exhausted battery could be the immediate cause of service loss; a shared fibre handoff could amplify the number affected; delayed landlord access could extend restoration. Naming only “power” hides the design and authority choices that turned an external event into a long outage.
Public documents do not provide Metrowireless’s incident command structure, supplier escalation times, spare levels, access agreements or mean-time-to-repair distribution. They also do not show a public severe-event exercise. That absence does not mean the practices are missing. It sets the verification agenda: inspect records, observe a drill and compare promised recovery with time-stamped performance.
The ultimate test is a controlled failure at the busiest credible time. Remove one radio relay, one site power source, one fibre handoff and one external route in separate exercises. Measure detection, diagnosis, dispatch, failover capacity, customer communication and full restoration. The results would turn today’s unresolved boundaries into evidence about actual authority and recovery speed.
What a serious resilience test must still prove
Metrowireless’s public record is substantial enough to avoid speculation about whether a network business exists. It establishes an authorized fixed-internet provider, point-to-point and point-to-multipoint wireless service, selected fibre offerings, a modeled coverage footprint, public internet number resources, visible route announcements, customer terms and local technical service. It also provides enough detail to show why a simple “wireless ISP” label is inadequate.
What the record does not establish is the physical graph behind those services. There is no verified inventory of active towers, rooftops and relays; no link diagram; no fibre route; no site-by-site power design; no disclosed access-sector or backhaul capacity; no busy-hour headroom; no proven diverse upstream path; no customer concentration by failure domain; and no restoration drill with timestamps. These are not decorative omissions. They are the variables that determine whether one local fault remains local.
The first verification task is a topology reconciliation. Begin with the 97 polygons in the dated KML, but do not treat them as assets. Associate active service areas with named access sectors, relays, aggregation points and handoffs. Record owner, location precision, power source, access authority and downstream dependency count. Archive retired areas and date every change. The output should make it possible to trace a customer service to every shared physical and logical component.
The second task is a capacity reconciliation. For each sector and transport link, compare hardware capability, configured rate, committed service load, busy-hour demand and failure-state demand. A service label belongs only in the customer column. The decisive question is whether the surviving path can carry priority and ordinary traffic after the largest single failure, and for how long.
The third is route and supplier verification. Confirm where AS263246’s external sessions terminate, which address families and prefixes they carry, and which physical circuits support them. Trace duct, pole, entrance and facility overlap. Require carriers to identify shared risk or explicitly state what they cannot verify. Then fail each path under load and observe whether inbound and outbound traffic move as intended.
The fourth is power and access testing. Measure actual autonomy at every shared site rather than quoting equipment nameplate values. Test batteries under load, generator start and connection, alarms, keys, landlord contacts and travel time. Include customer-side instructions because a healthy far end cannot serve an unpowered terminal. Repeat after battery aging and before the heaviest weather season.
The fifth is recovery rehearsal. Choose a radio misalignment, a relay outage, a fibre cut and an upstream failure. Timestamp detection, ownership decision, access, replacement, configuration, service validation and customer notice. Record where a supplier or property owner controls the clock. Use the result to set realistic restoration targets for ordinary faults and correlated regional events.
The public evidence supports a Medium network-evidence grade. The grade reflects strong confirmation of legal identity, authorized services, customer offers, modeled geography, number resources and observable routing, balanced against unresolved physical topology, installed-versus-usable capacity, route diversity, site power and failure recovery. It is not a performance score and does not predict an outage.
The most important watchpoints are concrete. A change to the public KML could indicate footprint revision but still needs site confirmation. A new or vanished route announcement could indicate a routing change but not its physical cause. A new infrastructure-sharing record could expand options but not prove use. Published changes to speed, contention or repair terms could alter customer exposure but not reveal available headroom. Independent measurements, incident records and verified physical maps would materially improve confidence.
At first light, the Santa Ana technician can solve one essential problem: whether this roof can see the network well enough to join it. The unresolved resilience question begins immediately afterward. What does that antenna depend on once it leaves the customer’s wall, how many other users share those dependencies, and who can restore each one when clear sky gives way to rain, lost power or a broken handoff? Until those answers are mapped and tested, the most honest view of Metrowireless is neither fragile nor invulnerable. It is a real regional network whose service edge is visible and whose concentration points remain largely private.