Baikal Teleport CJSC makes a local connectivity bill depend on upstream routes and field repair

Summary

• Baikal Teleport presents a credible current operating footprint: its website was updated in 2026, it says it serves more than 20,000 subscribers in more than 400 named locations, recent notices describe new fibre builds, and AS59616 was globally visible on 10 July 2026 with twelve IPv4 announcements.
• The network is not one uniform fibre system. Company notices in 2025 and 2026 repeatedly describe moving existing customers from radio links to newly built optical lines, while an official regional power document identifies a Baikal Teleport communications base station at Glazunovo. Reliability therefore depends on the particular access generation and route at each address.
• Public routing observations show two current upstreams, ER-Telecom Holding's AS9049 and Baikal-IX transit network AS57277. That is meaningful logical diversity, but no public evidence establishes physically separate exits, independent ducts, separate powered facilities or tested failover capacity.
• Baikal Teleport's own service messages expose the practical failure chain: a truck-damaged cable on Melnichny Trakt carried a four-to-five-hour repair estimate, electrical work in Granovshchina produced a one-to-two-hour estimate, and overnight network upgrades allowed planned interruptions. The value of the monthly connection consequently depends as much on spares, access rights and field crews as on advertised speed.

A small bill sits on a large physical chain

The most revealing number on Baikal Teleport's consumer broadband page is not the highest speed. It is the entry price. The company's home internet offer advertises optical access at up to 1 Gbps from 450 roubles a month, while another summary block on the same page presents up to 500 Mbps from 550 roubles. The variation is a useful warning against treating one headline as a universal tariff. Price, speed and connection cost are location-dependent across a footprint made up of a regional city, villages, cottage developments and hundreds of garden associations.

A customer sees one monthly debit. Behind it sit a drop cable or radio receiver, a router and power supply, a pole or other support, a distribution segment, an aggregation node, a backhaul path, an internet edge, an upstream handoff and a support organisation. Each part may have a different owner and a different restoration clock. The access line can survive while the upstream fails; the upstream can remain healthy while a lorry tears down a local span; the provider's equipment can stay energised while a customer's router goes dark. A low price does not simplify that chain.

It makes disciplined construction, sharing and repair more important because there is less revenue per line with which to absorb idle capacity and duplicated plant.

Baikal Teleport is a particularly useful case because its public history crosses several access generations. The company's account of its development says it began in 1999 with DECT radio access for telephony in Irkutsk and its suburbs, then became an early internet provider to the city's private-house sector. It now describes thousands of kilometres of optical line, more than 20,000 subscribers and service in more than 400 settlements or named localities. Recent notices continue to describe customers being moved from radio to fibre. The present network is therefore best understood as an expanding optical system laid over the geography and operating habits of an older wireless one, not as a greenfield fibre grid completed everywhere at once.

That distinction changes what a broadband bill buys. An address already reached by a modern optical distribution line may receive a high-capacity access medium, but it still depends on the route from that locality toward Irkutsk and on the power and switching facilities along the way. An address still served by radio adds line-of-sight, tower or rooftop access, interference and powered base-station dependencies. A newly converted settlement may have both technologies in transition, with technicians scheduling customer migrations while maintaining the legacy service until the last workable cutover.

The bill is local; the dependency chain is not.

The operator is active, even if every scale claim is not audited

The evidence supports treating Baikal Teleport as a current operator rather than a dormant registration. Its main site and service pages carry 2026 updates. Its subscriber portal remains available for account, service and payment functions. A February 2026 notice opened service orders in Ust-Baley, and a March 2026 notice said an optical network on Baikalsky Trakt was built and ready for connections. The company's routing number was visible to every full-feed RIPE RIS peer counted in a 10 July 2026 routing observation. Current operation is not inferred from one old commercial listing; several different public surfaces move at the same time.

The legal identity also joins up across those surfaces. Baikal Teleport's published company particulars give the closed joint-stock company name, tax identifier 3811060180, state registration number 1023801534261 and an Irkutsk legal address. The site identifies Yuri Pastukhov as general director and lists three communications licence numbers in its footer. The RIPE organisation record for AS59616 names Baikal Teleport CJSC and carries the same state registration number. These matches reduce the risk that a similarly named broadband brand has been confused with the network operator.

The company's scale statements still require qualification. Its claim of more than 20,000 subscribers is self-reported, as is the statement that it has built thousands of kilometres of fibre. The location selector on its website visibly contains a very long list of villages, settlements, neighbourhoods, cottage developments and gardening communities, so the claimed breadth is plausible. But a selectable place is not proof that every address there is serviceable, that all listed lines are active, or that all customers use fibre. The provider itself asks for an address before confirming terms.

A careful description is therefore that Baikal Teleport markets and supports a broad Irkutsk-centred service area with continuing builds, not that a uniform gigabit network blankets every one of more than 400 names.

There is no comparable problem with the basic operating conclusion. The official Russian accounting resource records a 2024 business with 228.979 million roubles of revenue, and route collectors see its traffic-facing identity in 2026. The website publishes a staffed support number, active offers and connection notices. None of these proves high availability, but together they provide strong evidence that the company is trading, adding plant and carrying internet routes.

The service area follows Irkutsk's outward roads

Baikal Teleport's geography is more specific than the word "regional" suggests. The centre of gravity is Irkutsk and the private-house belt around it, extending along named tracts that organise suburban growth: Baikalsky, Kachugsky, Alexandrovsky, Melnichny, Goloustnensky and others. The company's location selector includes Irkutsk, Shelekhov, Markova, Khomutovo, Urik, Ust-Kuda, Baklashi, Pivovarikha and many smaller named communities, as well as a dense field of SNT and DNT garden or dacha associations.

Recent build notices trace that geography at addressable scale. In January 2026 the company said an optical network in SNT Rosinka on Kachugsky Trakt was complete and ready for connections. The notice offered service up to 500 Mbps and said existing subscribers would be contacted about moving from radio to fibre. In March, a similar notice said fibre had been built for SNT Irkutyanin and SNT Stroitel on Baikalsky Trakt, again explicitly promising a radio-to-optical migration. A November 2025 notice described a completed optical line for Kytsigirovka and Nikolsk, but marketed speeds there at up to 100 Mbps rather than 500 Mbps.

Those differences matter more than a smooth coverage map. They show that build status, available tier and legacy technology vary by locality. They also show that the physical network is being extended in branches along roads and settlement clusters. A road-oriented network can be economical: one feeder can pass multiple communities, and sales in a compact association can justify a distribution build. It can also create common-path exposure. If several settlements share the same feeder along a tract, one cut before the branch point can affect them together even if each local distribution area is healthy.

The February 2026 Ust-Baley notice confirms order availability but does not identify access technology or a route. It should not be used to draw fibre where the operator did not say fibre. This is a recurring boundary in regional-network analysis: a commercial service notice establishes that the provider seeks or accepts customers in a place; it does not by itself disclose the cable path, radio site, split ratio, backhaul capacity or restoration design.

The footprint is thus both extensive and granular. It is extensive because the company lists hundreds of distinct places and continues to open new optical areas. It is granular because each tract and association can have a different build year, access medium, feeder, power source and customer density. For a resident deciding whether the monthly offer is dependable, the relevant map is not the full company footprint. It is the route from that house to the first aggregation point and then the route from that point to a functioning upstream.

Radio has not disappeared; it is being worked out of the network

Baikal Teleport's origin in radio access is not merely corporate history. The company's current expansion notices make the legacy visible by promising to contact existing customers for migration from radio channels to optical fibre. This is direct evidence of a mixed network in 2026. It also helps explain why one generic description such as "wireless ISP" or "fibre provider" is incomplete.

An official Irkutsk regional publication provides a rare physical anchor. A 2021 list of electricity consumers subject to special supply categorisation identifies a Baikal Teleport communications base station at Glazunovo in Irkutsk District. It associates the site with the 10 kV Karluk-Glazunovo line, a transformer point and a 0.22 kV connection. The record does not describe antennas, backhaul, customer count or backup power, and it should not be generalized to every radio site. It does establish that at least one named node was a powered field installation with a traceable electricity dependency.

Fibre removes some radio-specific failure modes. It does not need a clear Fresnel zone, is not degraded by the same interference mechanisms and can carry far more capacity over a passive span. Yet a fibre conversion introduces its own work. New cable needs poles, ducts or buried routes; closures and splitters need weatherproof placement; optical power budgets must accommodate distance and split; customer drops need to be installed; old radio customers need appointments and compatible premises equipment. A cut fibre has no graceful signal fade. It works until the glass is broken or the active equipment at one end fails.

The migration can also create a temporary operational double load. Technicians must keep the radio system usable for customers not yet moved while activating and troubleshooting the new optical system. Inventory may include radio customer units, optical network terminals, routers, connectors, splitters, cable, closures, power supplies and surge protection. Support staff must identify which access generation a caller has before giving useful advice. The same settlement name may contain both old and new service for a period, so an outage report without technology and address can be misleading.

Baikal Teleport markets fibre as the stable successor, and the move is directionally sensible. But "fibre" is a medium, not a resilience certificate. The public notices do not state whether the new tract lines form rings or terminate as branches, whether they use more than one road crossing, whether distribution nodes have alternate feeds, or how much spare capacity is held for a radio-to-fibre migration. The observable achievement is that optical plant has been built and customers are being invited to move. The recovery properties of that plant remain largely undisclosed.

The tariff reveals the economics of keeping the chain alive

The company's tariff communications show that physical maintenance reaches the bill. In November 2025 Baikal Teleport told customers that the price of its "port provision" service would become 60 roubles a month from 1 December. A companion tariff notice attributed broader price changes to inflation in customer and network equipment, components and consumables, higher operating expenses and a tax change. This was a marketing explanation, not a cost audit, but the named pressures are consistent with a regional operator that is simultaneously extending fibre and maintaining older access.

The 2024 financial statements available through Russia's state accounting resource put scale around that problem. They report revenue of 228.979 million roubles, up from 195.655 million in 2023; cost of sales of 148.713 million; gross profit of 80.266 million; management expense of 51.977 million; and operating profit of 28.289 million. Other expenses of 20.651 million and other charges left net profit at only 52,000 roubles, down from 12.042 million the previous year. Fixed assets rose from 65.634 million to 82.172 million roubles.

One year cannot establish the permanent economics of the company, and the statements do not separate broadband construction from other services. They do, however, show a growing revenue base, a substantial fixed-asset balance and almost no final profit cushion in 2024. The important conclusion is not that Baikal Teleport cannot invest; fixed assets did increase materially. It is that the margin available after all reported expenses can be much thinner than the gap between a household tariff and an advertised speed suggests.

Regional expansion intensifies that trade-off. A dense apartment building spreads one feeder and one technician visit across many paying units. A line along a tract to private houses and garden communities may cover longer distances, cross more third-party land or support structures, and generate fewer customers per kilometre. A one-day installation promise is commercially attractive, but it requires nearby stock, route knowledge and available labour. A 24-hour support line is useful, but the person answering cannot splice glass, replace a damaged drop or reach a snowbound cabinet.

The provider can improve the equation through clustering, reused routes, local peering, remote diagnosis, customer self-service and migration from maintenance-heavy radio links. It can also sell television, video surveillance, telephony and business service over the same access. Those efficiencies help explain how a relatively low monthly charge can support a broad footprint. They do not remove the need to finance route diversity, reserve equipment and the occasional repair that produces no new revenue but preserves every affected line.

AS59616 has two visible ways out, not proof of two separate roads

At the internet edge, Baikal Teleport operates AS59616. The RIPE registration lists routing policies with ER-Telecom Holding, Baikal-IX networks and several local systems. On 10 July 2026, RIPE's routing status showed the autonomous system announced, visible to all 327 IPv4 full-feed peers in that observation and originating 2,304 IPv4 addresses. Its announced-prefix view returned twelve routes, including the aggregate 91.246.0.0/22 and blocks in 193.106.0.0/23 and 193.169.0.0/23, along with more-specific /24 routes.

The current upstream picture is better than a single-transit design. RIPE's neighbour observation showed two unique left-side neighbours: AS9049 and AS57277. A separate live network summary identifies them as JSC ER-Telecom Holding and BAIKAL-IX, operated by LLC Zero Kilometer. Both carried IPv4 reachability; neither showed IPv6. The RIPE routing view likewise reported no visible IPv6 space, even though an IPv6 route object exists for 2a05:a7c6::/32. That difference between registration and observation is another reason to distinguish configured intent from live service.

The local exchange relationship has real value. The Baikal-IX site provides a regional peering platform and looking glass, while the RIPE policy for its AS48586 explicitly includes Baikal Teleport's AS59616. Local exchange can keep traffic between participating Irkutsk networks and content systems closer to users, reduce paid long-haul transit and shorten some paths. AS57277 is more than an exchange route server: its current policy describes a transit network with several upstreams and lists AS59616 as a customer. That gives Baikal Teleport a second logical path to the wider internet alongside ER-Telecom.

Logical diversity is necessary but not sufficient. Two BGP neighbours can terminate on the same router, enter the same building, use fibres in the same duct, cross the same bridge or rely on the same electrical room. AS57277 and the Baikal exchange are local to Irkutsk; a connection to them may improve route choice without creating a geographically independent exit from every suburban feeder. ER-Telecom may deliver on a different physical path, but no public source here establishes the handoff locations or cable routes. A routing table can prove that two networks announce a way to reach AS59616.

It cannot prove that an excavator, building fire or utility failure cannot remove both handoffs at once.

There is also a route-security gap worth separating from availability. RIPE's RPKI validation response for 91.246.0.0/22 returned "unknown" because no validating route-origin authorisation was found. That is not the same as an invalid route, and it does not make current service unreachable. It means origin validation does not add a cryptographic authorization signal for that route in the observed state. Creating and maintaining valid ROAs would reduce one class of routing error or hijack exposure, but it would not protect a cable or power feed.

The responsible resilience conclusion is therefore mixed. Baikal Teleport is not publicly visible as a one-upstream stub: it has two current upstream neighbours and a local exchange policy. That is a meaningful strength. The physical independence, failover testing, contracted capacity and congestion headroom behind those neighbours are not public. They should remain unverified rather than being assumed from the number two.

A truck on Melnichny Trakt is the clearest failure test

The most useful outage evidence comes from the operator's own customer messages. In a post preserved on its public Telegram channel, Baikal Teleport warned customers on Melnichny Trakt that access could be limited after a cable was damaged by freight transport. It said a crew was working on the fault and estimated restoration in four to five hours. Another message in the same public sequence told customers in Granovshchina that electrical works had caused a technical failure, with a one-to-two-hour repair estimate. The company also announced overnight network-modernisation windows during which interruptions were possible.

These notices expose more about the operating network than a speed claim does. Melnichny Trakt had a cable positioned where road activity could damage it, directly or through a struck support. The event affected a named corridor rather than the whole autonomous system, which points toward an access or aggregation segment. The four-to-five-hour estimate suggests the response required more than a remote configuration change: a crew had to locate, access and repair the physical fault. The message does not say how many customers lost service, whether the cable was optical, or whether an alternate path carried some traffic.

Those facts remain open.

The Granovshchina notice shows a different boundary. Work by electricians affected telecommunications service. That can happen when power to active equipment is interrupted, when utility work damages communications plant, or when safe access requires equipment shutdown. The post does not identify the exact mechanism. It nevertheless confirms that the service can be exposed to another trade's work and that recovery depends on coordination outside the network operations desk.

For customers, four hours and one hour are not abstract availability percentages. Four to five hours can consume a remote worker's afternoon, stop card authorization at a small shop, interrupt cloud cameras or push a household onto mobile data. One to two hours can still outlast a small uninterruptible power supply or a scheduled online lesson. The effect varies by time of day, alternate connectivity and customer activity, but the locality-specific repair clock is the actual product during failure.

The notices also show good operational behaviour in limited form. The company named affected areas, described a cause at a useful level, acknowledged that a crew was working and gave an estimate. Public communication does not prove that the estimate was met, nor does it substitute for redundancy. It does reduce uncertainty for customers deciding whether to wait, travel or switch to a backup. A mature follow-up would add restoration time, scope and whether permanent protection was installed, but no such post was found alongside the captured notices.

Melnichny Trakt should consequently be treated as a real resilience test, not an anecdote. It demonstrates a plausible common-path failure and the importance of a local repair team. It also defines the evidence needed for a stronger design: an alternate route around the damaged segment, automatic or rapid traffic transfer, tested capacity on that route, accessible spare cable and closures, and a crew able to work safely at the hour and in the weather presented.

Power turns every passive line back into an active service

Optical distribution can be passive between endpoints, but broadband service is never free of electricity. The customer router and optical terminal need power. So do aggregation switches, optical line terminals, radio base stations, edge routers and upstream handoffs. The Glazunovo record makes one such dependency concrete by tying a Baikal Teleport base station to a named 10 kV line and transformer connection. It says nothing about battery runtime or a generator.

Irkutsk's regional environment makes that omission important. In April 2025, wet snow and winds reported at up to 35 metres per second interrupted electricity in 394 settlements across Irkutsk Oblast. That report is not evidence of a Baikal Teleport outage and covers a much wider area than the operator's core footprint. It is evidence that weather can produce geographically broad electricity failures in the same region. In October 2025, snow and wind cut supply to Markova and eleven garden settlements in Irkutsk District, affecting 2,052 private homes. Markova and garden associations are precisely the kind of localities that appear in Baikal Teleport's service selector.

The regional system operator has described Irkutsk Oblast as an area with high electricity-supply disruption risk in 2025-2027. Its winter exercise included a scenario in which a line fault removed power from more than 18,000 people at minus 33 degrees Celsius. The exercise demonstrates planning, not a forecast that Baikal Teleport will fail. It sets a credible design condition for any network serving private homes in the region: commercial power may be absent during cold weather, road conditions may slow access, and restoration priorities belong to the electricity operator.

Backup power must be specified node by node. A battery that supports a cabinet for fifteen minutes is different from one sized for several hours. A generator at a central office does not energise an unprotected field switch. A radio site may remain up while homes below it have no power to run receivers and Wi-Fi. Conversely, households with their own generators may still have no service if an upstream aggregation point is dark. Claims such as "the network has backup" are therefore limited public evidence without runtime, load, maintenance and refuelling detail.

No public Baikal Teleport source reviewed here provides that detail. There is no published inventory of batteries, generators, dual utility feeds, alarm telemetry or winter fuel plans. Nor is there evidence that two upstream handoffs sit in independently powered facilities. This does not mean the measures are absent. It means power resilience cannot be credited beyond the observed fact that the operator restores local faults and offers around-the-clock support.

The field crew is part of capacity

Network capacity is usually expressed in bits per second. For a regional provider, repair capacity also has a human unit: how many qualified people can be dispatched, how far they must travel, what equipment they carry and how many simultaneous faults they can absorb. Baikal Teleport's footprint makes this especially material. A technician may need to move between Irkutsk, an outlying village and a garden association along a separate tract, while another team continues scheduled installations.

The company advertises 24-hour technical support, but call availability and field availability are different. Its employment page invites applicants without publishing a permanent technical roster. A public post from the company's channel advertised an engineer for access-network subscriber service, which is a direct sign that this work exists as a defined local role. The hiring notice does not establish a shortage, team size or vacancy duration. It supports a narrower point: the service depends on retained access-network expertise, not only outsourced construction or a remote call centre.

The operator's one-day connection promise likewise implies local logistics. To connect a private house quickly, a team needs an approved route, pole or property access, sufficient drop cable, connectors, premises equipment, a configured service profile and a vehicle. A radio installation may require safe mounting and alignment; a fibre installation may require splicing or pre-connectorised drop work and optical testing. Weather, locked sites, landowner permissions and the condition of supporting poles can turn a simple order into a field problem.

During an outage, installations compete with restoration unless the workforce is separated or contractors can be called in. A truck-damaged feeder can require traffic control, replacement span, closure work, testing and customer confirmation. Multiple power-related faults after a storm can scatter crews across a wide area. A network with spare electronic capacity can still recover slowly if it lacks people, vehicles, access keys or compatible components. Conversely, a compact, experienced crew with good route records can restore a modest network faster than its scale might suggest.

Baikal Teleport does not publish mean time to repair, crew count, depot locations, spares policy or contractor arrangements. Customer-review sites contain comments about installation and support, but those reports are self-selected and cannot establish organisation-wide response. The Melnichny and Granovshchina notices are stronger because they name incidents and estimates. They show that the company dispatches teams; they do not show how the organisation performs under two or three simultaneous tract failures.

For buyers who need business continuity, the provider's business internet page is notable because it offers the organisation of backup communications channels. That is the right product concept. The useful follow-up is whether the backup uses a separate physical carrier and route, how failover is tested, which equipment remains powered, and what service target applies to repair. A second logical circuit delivered over the same damaged tract is an accounting duplicate, not a recovery path.

Advertised speed is installed potential, not guaranteed usable capacity

Baikal Teleport advertises residential optical tiers up to 500 Mbps or 1 Gbps depending on the page component and location, and business service up to 1 Gbps. Recent locality notices vary: Rosinka and the two Baikalsky Trakt associations were marketed at up to 500 Mbps, while Kytsigirovka and Nikolsk were offered up to 100 Mbps. That variation is evidence of address-specific engineering or commercial choices, not an inconsistency that can be averaged away.

An optical port capable of a gigabit is installed potential. Usable capacity depends on the access design, optical budget, customer equipment, shared segment, aggregation uplink, transit and traffic pattern. A 500 Mbps tariff can perform well when only a fraction of customers are busy at once and the feeder has headroom. It can also fall short at the evening peak if too many high-rate lines share a constrained aggregation path. Public routing data shows reachability and neighbours, not the number or size of physical ports between Baikal Teleport and those neighbours.

Failover capacity is the harder test. If normal traffic is divided across AS9049 and AS57277, losing one path places more demand on the survivor. The surviving contract and port must be able to carry the shifted load, and routing policy must move traffic quickly enough. A second BGP session that has only enough capacity for ordinary background traffic may preserve reachability while producing severe congestion. No public utilization graphs or contracted capacities were found for AS59616.

The access migration adds another capacity question. When radio customers are shifted to fibre, their potential peak demand rises sharply. The fibre line may remove a last-mile radio bottleneck but expose a shared feeder or upstream limit. The company may have upgraded those layers in parallel; the build notices do not say. Evidence that would settle the question includes optical line terminal port design, split ratios, aggregation capacities, busy-hour utilization before and after migration, and demonstrated throughput during a single-upstream condition.

Customer premises are part of the result. The company sells or supports modern routers, but Wi-Fi conditions inside a house can make a sound fibre service appear slow. Baikal Teleport's own speed-test guidance exists because device, wireless and measurement conditions matter. This is not a reason to dismiss complaints. It is a reason to separate access-line performance from in-home distribution and to test by cable when diagnosing a tariff-speed gap.

The correct interpretation is consequently local and conditional. The network has demonstrated the ability to sell high-speed optical access in named communities, and current routing supports a substantial active edge. There is not enough public evidence to claim that every advertised tier is continuously usable at every address, or that full rates survive a route failure. Capacity should be credited as installed commercial capability, with resilience and busy-hour performance still to be verified.

Not every visible outage begins inside Baikal Teleport

The ownership boundary matters when service degrades. A customer generally calls the retail provider no matter where the fault lies. Yet Baikal Teleport may control only part of the path. It controls its access equipment, configuration and customer support. A pole owner, road contractor or electricity company can control the conditions around the local line. ER-Telecom and Baikal-IX transit control their networks beyond the handoff. State-level traffic controls and remote platforms can affect particular services while the physical access remains healthy.

Baikal Teleport made that distinction in a February 2026 notice about reduced Telegram performance, saying the condition did not arise from its own network or equipment. An earlier public-channel post attributed widespread instability across fixed and mobile networks to external gateway testing. These statements are the operator's account and are not independent technical post-mortems. They illustrate why "the internet is down" must be decomposed before assigning a repair action.

The diagnostic sequence should begin at the premises: is the terminal powered, does the local link show carrier, and do multiple devices fail? It should then move through the access area: are neighbours on the same line affected, and is the radio or fibre node reachable? From there the operator can test aggregation, DNS, route availability and upstream performance. A failure affecting one application but not general IP connectivity calls for a different response from a dark optical terminal or an absent BGP route.

This division does not remove the retail provider's responsibility to communicate. The subscriber cannot negotiate with an upstream carrier or inspect a field cabinet. Baikal Teleport's value lies partly in recognizing the fault domain, escalating to the right owner and maintaining updates. The Melnichny post named physical damage; the Granovshchina post named electrical work; the 2026 notice named an external service restriction. Those distinctions help, even though public restoration reports would make them more useful.

For business customers, the boundary should be reflected in contract design. A backup that uses a different access technology but the same upstream may survive a pole cut and fail during an upstream incident. A second upstream on the same access route has the opposite profile. A mobile backup may avoid the fixed line but remain exposed to regional power or traffic restrictions. There is no single independent backup; independence must be tested against the actual failure the customer cannot tolerate.

Market signals support activity, not a reliability guarantee

Several unofficial signals reinforce the current operating picture. A 2GIS listing updated in 2025 markets optical service to private homes and dachas and shows a large volume of user interaction. A provider-comparison site carries customer reviews dated into 2026, including comments on installation, speed and support. A separate outage-reporting page recorded a concentrated group of user reports on 27 May 2026 saying service was absent or intermittent.

These signals have narrow uses. Recent reviews and outage reports suggest that people still identify themselves as Baikal Teleport customers and interact with a live service. They cannot prove subscriber count, average speed or a network-wide incident. Review platforms may verify some submissions but remain self-selected; satisfied and dissatisfied customers post at different rates. An outage counter can collect duplicate, mistaken or geographically unrelated reports, and it usually cannot identify the fault domain.

The May cluster is therefore not used here as proof of a Baikal Teleport-wide outage or as evidence of a specific cause. It is a reason to seek a more authoritative record: an operator status notice, route change, monitoring series or restoration statement. No matching public operator explanation was found. The absence of a public explanation does not mean there was none through customer channels. It means the cluster should remain an unofficial signal.

The same discipline applies to positive comments. A customer saying that service has been stable for two years can provide texture about one address. It cannot establish route diversity or predict performance at another association on a different feeder. Baikal Teleport's own incident notices, live route visibility and named fibre builds carry more weight because they can be tied to specific infrastructure or time. The unofficial material is useful mainly where it aligns with those stronger facts or identifies a question worth testing.

Who is affected when the chain breaks

Baikal Teleport's service mix broadens the effect of an outage beyond web browsing. The company sells home internet, interactive television, video surveillance, telephony and business access. Its own claims put more than 20,000 subscribers on the network. That figure is not independently audited, but it indicates the order of dependency the company says it carries.

For a private-house customer, the connection may be the main fixed alternative to mobile service. A tract-level cable cut can remove work calls, school access, messaging, entertainment and cloud services at once. Interactive television may fail with the internet because it rides the same access. Cloud-connected cameras can stop sending live video or recordings off site, although local storage may continue if the camera and premises retain power. A power outage can disable all customer equipment even if the provider's line is intact.

For a small business, the consequences can be more immediate: payment terminals, inventory systems, bookings, remote access, voice and security monitoring may share one connection. Baikal Teleport's business page explicitly offers reserve-channel organisation, which acknowledges this dependency. A customer accepting that offer should ask whether the second channel has independent access, aggregation, power and upstream paths, and whether the router changes over automatically.

The geography can make mobile substitution uncertain. Some of the company's target communities sit outside dense urban coverage or have households that adopted fixed service precisely because radio or mobile alternatives were limited public evidence. Irkutsk Oblast's digital-development programme said 198 settlements with populations between two and 500 lacked both mobile and internet access when the programme was framed, and identified fibre expansion as a priority. That figure does not define Baikal Teleport's service area, but it explains the regional policy context: a local fixed provider may serve places where alternate connectivity is not assured.

The harm is not identical for every failure. A cut on a distribution branch may affect one association. Loss of a shared tract feeder can affect several. Failure of an Irkutsk aggregation point or both upstream exits could be much broader. The public record does not expose enough topology to estimate each blast radius. That is itself an important customer and procurement question: which localities share each single point, and what can be rerouted when it fails?

What a credible resilience demonstration would show

Baikal Teleport has already supplied several pieces of a credible operating case: named fibre builds, evidence of legacy radio migration, a current autonomous system, two observed upstreams, a regional exchange relationship, around-the-clock support and examples of dispatched repair. The missing pieces are not more slogans. They are tests of independence and recovery.

For upstream diversity, the useful evidence would identify handoff facilities for AS9049 and AS57277, the physical carriers and entrance routes, port capacities and the result of a controlled single-upstream failover under busy load. It would show whether DNS, authentication, monitoring and customer address assignments continue to work after the route change. It would also clarify whether the additional relationships listed in RIPE policy are active peerings, historical configurations or available contingencies.

For optical access, a meaningful topology account would distinguish rings from branches and identify where separate-looking paths share poles, ducts, bridges or road crossings. It would state which tract feeders can be rerouted, how many customers sit behind major aggregation points, and whether critical spares are stored locally. The Melnichny incident supplies a practical benchmark: could an alternate route have carried the affected customers for the four-to-five-hour physical repair, and if not, what route protection followed?

For radio, the operator would need to identify remaining service areas, base-station overlap, independent backhaul and power, spare radios and customer units, and the timetable for migrations where fibre is under construction. Two radio sites that share one backhaul or one utility line do not create end-to-end redundancy. Fibre migration should reduce some failure modes without silently concentrating customers on one new feeder.

For power, evidence should include battery runtime at edge, aggregation and radio sites under real load; generator coverage and connection points; fuel and cold-start testing; alarm visibility; and the restoration plan for nodes on high-risk utility circuits. The customer layer should be addressed honestly: the provider can keep its network alive but cannot power every home router. Publishing equipment consumption and compatible backup options would let users decide what continuity they can buy themselves.

For field repair, useful measures include staffed response hours by area, average and upper-bound dispatch time, concurrent incident capacity, winter access arrangements, contractor call-out terms and stock of cable, closures, optics, power supplies and customer equipment. Public incident communication should close the loop with restoration time and cause. That would turn one-way estimates into evidence of performance.

None of these disclosures needs to expose sensitive network coordinates or customer data. Aggregated topology classes, tested failover results and restoration measures can show resilience without publishing a map of vulnerable assets. The aim is to distinguish a network that has more than one item from a network that can continue service when one of those items is deliberately removed.

Assessment: current service is credible; independence is the open question

Baikal Teleport's operating-status hypothesis can be upgraded. The company has a live 2026 web and account presence, current fibre-construction notices, an active support surface, a substantial 2024 revenue base and a globally visible routing identity with two observed upstreams. The public evidence is not thin enough to justify describing current operation as uncertain.

The resilience assessment remains more restrained. The company has moved from its radio origins toward a large optical footprint, but its own notices show that the conversion is incomplete. It has two visible upstream networks, but the physical paths and failover headroom are not disclosed. It operates or has operated powered field sites, but backup-power duration is not public. It dispatches crews to cable and electricity-related failures, but crew depth, spares and restoration performance across simultaneous incidents are unknown.

The economics explain why those questions matter. Revenue and fixed assets grew in 2024, yet final profit was almost erased by other expenses and charges. The company increased prices in late 2025 while naming equipment, materials and operating costs. A broad private-house network earns its return one modest bill at a time, across long routes that can be damaged by vehicles, utility work and weather. Redundant plant is valuable precisely because it is underused most days, which also makes it difficult to fund and tempting to infer where it has not been demonstrated.

The strongest conclusion is therefore not that Baikal Teleport is fragile, nor that its fibre build guarantees resilience. It is that the operator has assembled a credible, growing regional network whose customer value is determined at the junction between local plant and recoverability. The monthly tariff buys access to that system. Whether it also buys continuity depends on address-specific technology, feeder topology, powered nodes, upstream physical diversity and the technician who reaches the break.

That is why the Melnichny Trakt repair is more instructive than a gigabit headline. A freight vehicle damaged a cable; customers faced a four-to-five-hour estimate; a local crew worked the problem. Every regional network eventually meets some version of that event. Baikal Teleport's next level of evidence would show which customers can route around it, how long the powered network can wait, and how much capacity remains when one path is gone.