OM SHIV SAI INTERNET SERVICE (OPC) PRIVATE LIMITED makes a local connectivity bill depend on upstream routes and field repair

Summary

• The operating company is well supported, but its shortened name needs correction. India's Department of Telecommunications, TRAI, APNIC, the company's own contact page and both major mobile app stores identify OM SHIV SAI INTERNET SERVICE (OPC) PRIVATE LIMITED. DoT lists a Category C ISP authorisation for Kalyan, signed and effective on 17 November 2021.
• Subscriber reporting supports a real and growing access business. TRAI listed 900 broadband subscriptions in each available quarterly snapshot from September 2022 through December 2023, then 1,200 in June 2024, 3,060 in September 2024, 4,000 in December 2024 and 3,900 in March 2025; a later TRAI dashboard snapshot lists 4,200. The movement is meaningful, although the regulator says these statistics rely mainly on provider submissions.
• AS149214 was active and globally visible on 10 July 2026. It originated four IPv4 /24s and one IPv6 /48 with valid route-origin authorisations. Two IPv4 blocks are registered to Om Shiv Sai; two newer /24s are registered to another Maharashtra network and therefore demonstrate routing responsibility, not Om Shiv Sai ownership.
• PeeringDB declares operational 10 Gbps and 30 Gbps ports at DE-CIX Mumbai and Extreme IX Mumbai, plus six interconnection facilities in Mumbai, Thane and Vasai. This is credible logical and metro-site diversity. It does not establish two physically independent paths from a customer's street, two full-transit contracts, spare capacity after a failure or diverse building entrances.
• The company's customer app supports billing, usage and complaint handling, which makes local support labour part of the product. Public evidence does not disclose crew count, shift coverage, spares, fibre or wireless mix, pole and duct rights, node backup runtime, mean restoration time, route maps or major-outage performance. The final network evidence grade is Medium for current operation and Weak for physical resilience.

The first correction is the name

The shorthand name omits a word that appears consistently in authoritative records. The Department of Telecommunications' ISP authorisation list current to February 2026 names OM SHIV SAI INTERNET SERVICE (OPC) PVT. LTD, licence DS-11/199/2021-DS-III. It classifies the authorisation as Category C, gives Kalyan as the service area, names Sanjay Vishnu Sawant as director, and records 17 November 2021 as the signing and effective date. The registered office in that row is at Shri Prashta Complex, Nallasopara West, in Maharashtra's Thane district.

The company's own contact page supplies the same legal name, a matching licence number and the corporate identification number U72900MH2020OPC347747. APNIC's AS149214 record uses the same Om Shiv Sai name and the same Shri Prashta address. Google Play's OSS Internet listing names the developer as OM SHIV SAI INTERNET SERVICE (OPC) PRIVATE LIMITED and repeats that address. Apple's OSS Internet listing names the same provider. Five surfaces with different functions converge on one identity.

That convergence matters because “Shiv Sai Internet Service” is not merely an editorial abbreviation. Dropping “OM” weakens the join between the reader-facing company, the licence, the ASN, the app developer and the legal entity. This article retains the existing directory link, but readers should understand that the public evidence supports Om Shiv Sai as the complete name. The title and entity display should ultimately follow the official form.

The identity correction does not imply a new company, a rename event or a corporate relationship. It is a correction to the label attached to the existing directory company. Nor does the evidence support a global region. The licensing boundary is Kalyan; the office, customer contacts, facilities and internet exchanges are in Maharashtra; and the registered country for the network resources is India. The regional-ISP category is defensible. “Global” is not a defensible description of the service area.

There is one further reason to establish identity before discussing infrastructure. Small broadband networks often appear under several layers of naming: a legal company, a retail brand, an autonomous-system name, a billing portal and local cable operators who install or maintain the final connection. Here, the public site uses “Om Shiv Sai Internet Service,” the app uses “OSS Internet,” and the ASN uses OMSHIVSAI-AS-IN. These are useful identity signals, but they do not tell us who owns each pole, cable, radio, building entry or long-haul circuit. A customer can pay one brand while depending on assets controlled by several parties.

A regulator-reported subscriber base turns the company from a registration into an operator

The strongest operating evidence is not the website's claim to be a leading provider. It is the sequence in TRAI's ISP subscriber annexes. The July-September 2022 performance report listed Om Shiv Sai with 900 broadband subscribers. The October-December 2022 report again listed 900. The April-June 2023 report, July-September 2023 report and October-December 2023 report each retained the 900 figure.

The series then changed sharply. TRAI's April-June 2024 report listed 1,200 subscriptions. The July-September 2024 report listed 3,060. The October-December 2024 report listed 4,000, and the January-March 2025 report listed 3,900. A later TRAI telecom dashboard compilation presents a 4,200 count for Om Shiv Sai.

The exact path between those snapshots should not be smoothed into a continuous growth curve. Quarterly counts are reported at quarter end, can be revised, and TRAI explicitly warns that its statistics are mainly reliant on data obtained from service providers. A change from 4,000 to 3,900 can reflect churn, reporting scope, cleanup or real customer loss; the public table alone does not decide which. The later 4,200 figure also does not reveal how many accounts are active, paid, residential, business, wired or fixed wireless at a given moment.

Even with those caveats, the series is far stronger than a static licence. A company that repeatedly appears with a non-zero customer base, changes scale, maintains a customer app, originates internet routes and joins exchanges has crossed the evidence threshold from “registered possibility” to “operating regional ISP.” The appropriate downgrade is no longer about whether Om Shiv Sai exists. It is about what kind of access network it runs, where that network ends, and how well the physical dependencies survive faults.

The growth also creates an operational question. Holding at 900 for more than a year and then rising above 3,000 within two quarters implies a material change in subscriber capture, reporting or both. If the increase was physical expansion, technicians had to install customer equipment, extend drops, activate ports and absorb a larger complaint load. If it was mainly a reporting or commercial consolidation, responsibility for existing access plant may have moved under the Om Shiv Sai service count without every asset changing hands. Public evidence does not distinguish those scenarios.

Either way, a 4,000-account network is large enough for shared failure domains to matter and small enough that field labour and spares can remain concentrated. Ten failed customer drops are a queue. A feeder cut, powered-node outage or lost upstream can affect hundreds at once. The subscriber count therefore supports the Regional ISP economics and Local support labour topics, but it cannot by itself prove scale economies, staffing depth or resilience.

Kalyan is the legal service boundary; Nallasopara and the Mumbai edge show the practical centre of gravity

DoT's Category C row defines the licensed service area as Kalyan. The registered-office evidence and customer-facing contact surfaces point to Nallasopara West, north of central Mumbai. PeeringDB's AS149214 profile declares facilities in Mumbai, Thane and Vasai. This cluster makes Maharashtra, not “Global,” the useful geographic frame.

The distinction between a licensing label and an actual access footprint matters. Kalyan is a licence service area, not a cable map. Nallasopara is an office and contact location, not proof that every nearby building is serviceable. A presence at a Mumbai facility is an interconnection claim, not proof that Om Shiv Sai owns a continuous fibre route from that facility to each subscriber. The website offers no address-level coverage map that can be trusted as evidence. Indeed, its home page contains a “San Diego” coverage heading and unrelated template testimonials, so those elements cannot be treated as Maharashtra network facts.

The company's plans page lists six consumer tiers from 100 Mbps to 300 Mbps, with 30-day prices from Rs 500 to Rs 1,750 and longer terms. Its services page advertises residential and business connectivity. Those pages show the retail proposition. They do not identify the access technology used at each address. The terms refer broadly to broadband, Wi-Fi and related services, while the plausible access mix includes fibre and fixed wireless. Without an engineering page, equipment filing, installation specification or address-level disclosure, the responsible description is “regional broadband access network,” not universal fibre-to-the-home.

This is an important restraint. A 300 Mbps plan is compatible with several access designs. It could be delivered over passive optical fibre, active Ethernet, a local cable system, a short wireless link, or some combination across the footprint. The physical failure modes differ. Passive optical distribution can keep field nodes unpowered but still depends on an optical line terminal and customer optical terminal. Active Ethernet introduces powered switches in the path. Fixed wireless introduces tower, line-of-sight, spectrum and weather dependencies. Hybrid service areas can contain all three.

The available evidence therefore supports the presence of an access business without supporting a uniform asset inventory. It does not justify depicting a branded cabinet, tower or fibre ring as fact. A representative physical scene would show a Maharashtra field technician working on an unbranded neighbourhood distribution point, with both aerial communications plant and local power visible, because the exact access medium and ownership boundary remain unverified.

A low monthly price purchases a chain of shared dependencies

Om Shiv Sai's plan table is commercially revealing even if it is not an engineering specification. At the listed monthly rates, the operator cannot reserve a dedicated 100, 200 or 300 Mbps end-to-end circuit for every household at all times. Consumer broadband economics depend on statistical sharing: customers peak at different times, aggregation links carry many accounts, popular content is reached efficiently through peering, and field visits are grouped across a compact service area.

That model is normal. It is also why installed capacity and usable capacity are different. The speed printed beside a plan is a retail ceiling or service commitment. It is not the sum of dedicated capacity built through every aggregation link. A declared 30 Gbps exchange port says how fast that interface may run; it does not state average traffic, peak load, committed capacity, burst terms, route-server reach, or what remains if another port or transport circuit fails. PeeringDB's entity-maintained 50-100 Gbps traffic band is a broad declaration, not a meter reading.

The first dependency sits inside the customer premises. A router, optical terminal, radio or cable modem needs local power and a sound drop. The customer app may show an active account even when that equipment has lost electricity. A battery-backed phone can make a mobile network seem more resilient during a household outage, but fixed broadband equipment commonly stops when the socket goes dark unless the customer provides backup. No public Om Shiv Sai page specifies supplied battery runtime or customer power guidance.

The second dependency is the neighbourhood access path. An aerial drop can be damaged by construction, vegetation, vehicle contact or a support failure. A buried path can be cut during excavation or water works. A wireless link can be obstructed or lose power at its radio site. An apartment connection can fail at a building switch, riser or shared equipment room. The provider may own the electronics while a landlord, utility, local operator or road authority controls the support structure and access window.

The third dependency is aggregation. Many local links converge on switches, optical equipment or radio backhaul. Concentration is economical because it reduces equipment and transport cost per subscriber. It also creates a common failure domain. A single failed uplink, power supply or aggregation chassis can turn many individual “internet down” complaints into one infrastructure incident. Public records do not disclose Om Shiv Sai's node count, port reserve, ring design or spare chassis stock.

The fourth dependency is the metro route to interconnection. Om Shiv Sai's edge is visibly present at two exchanges and several facilities, but traffic must reach those sites. Two exchange ports in Mumbai do not automatically mean two ducts from Nallasopara, two providers, two bridge crossings, two building entrances or two independent power systems. Both ports can ride the same access tail or meet at the same intermediate aggregation point. Logical diversity becomes physical resilience only when shared-risk groups are known and separated.

The fifth dependency is upstream reachability. Public route collectors consistently identify AS137085 as the strongest provider-side neighbour for AS149214. Some commercial datasets list additional possible upstreams or peers, but those classifications vary with collector position and route-server visibility. A network can have many observed adjacencies while relying on one provider for a full internet table. The public evidence does not establish two contracted full-transit providers that can each carry the customer load independently.

The sixth dependency is human recovery. When remote diagnosis cannot restore service, someone must find the failed span, gain access, test light or signal, replace equipment, splice or reterminate cable, and close the complaint accurately. The OSS Internet app lets customers register and track complaints, pay bills, view usage and inspect plan details. That is a credible operational surface. It does not reveal how many technicians are on shift, how tickets are prioritised, whether the operator or a local contractor owns the job, or how long a multi-customer fault takes to repair.

AS149214 is active, dual-stack and better documented than the last mile

The internet edge is the clearest technical part of the company. APNIC registered AS149214 in January 2022 under the name OMSHIVSAI-AS-IN. The record was active and had recent contact maintenance when reviewed. RIPEstat's AS overview marked the ASN announced on 10 July 2026. Its routing-status endpoint showed 326 of 327 IPv4 collector peers and all 321 IPv6 peers seeing the network.

RIPEstat's announced-prefix record showed four IPv4 /24s and one IPv6 /48 in the current set: 103.178.253.0/24, 103.187.123.0/24, 163.128.164.0/24, 163.128.165.0/24 and 2001:df0:44c0::/48. The first two IPv4 blocks and the IPv6 block have APNIC records naming Om Shiv Sai. The 103.178.253.0/24 record dates from January 2022, the 103.187.123.0/24 record from May 2022, and the IPv6 /48 record from January 2022.

The other two /24s require more careful language. APNIC's 163.128.164.0/23 record names BOISAR NETWORK INTERNET OPC PRIVATE LIMITED, not Om Shiv Sai. AS149214 originated the two /24 components on 10 July 2026, and each had a valid route-origin authorisation for AS149214. This supports current routing responsibility. It does not transfer registry ownership to Om Shiv Sai, prove a corporate relationship, or show the commercial terms under which the routes are carried.

All five observed prefixes returned valid route-origin results. RIPEstat's validation records for 103.178.253.0/24, 103.187.123.0/24 and 2001:df0:44c0::/48 match AS149214. The two Boisar-registered /24s also validated for AS149214. That is good routing hygiene: networks enforcing route-origin validation have cryptographic authority to accept these origins.

RPKI does not make the service physically redundant. It answers a narrower question: is this ASN authorised to originate this prefix? It cannot keep an access cable intact, provide electricity, prevent congestion or dispatch a technician. Likewise, a large IPv6 address space is not large traffic capacity. One /48 can number an enormous set of endpoints while travelling through a single constrained link.

The current routes also show interruptions within the two-week RIPEstat observation window. The announced-prefix timelines contain gaps around 6-9 July 2026 before all five prefixes appear again on 10 July. A route-collector gap can result from an origin withdrawal, upstream propagation, collector visibility or data timing; it is not enough to declare a customer outage. It is nevertheless a monitoring signal worth preserving.

The evidence that would settle impact is a timestamped operator incident notice, customer measurement from the service area, exchange-port telemetry, or a route-collector comparison showing a common withdrawal across independent collectors.

Two exchanges improve the edge without proving two ways out

PeeringDB's entity-maintained record is specific. It declares one operational 10 Gbps dual-stack port at DE-CIX Mumbai and one operational 30 Gbps dual-stack port at Extreme IX Mumbai. It also declares presence at Tata Communications Mumbai, Bharti Airtel Mumbai, Equinix MB1, STT Mumbai 2, Extreme Thane and Ishan Technologies Vasai. Extreme IX's technical entity list independently includes “Om Shiv Sai Internet Service” and AS149214.

This is meaningful infrastructure evidence. Exchange participation gives the network a place to receive routes and exchange traffic without sending every packet through paid transit. Two exchanges reduce dependence on one exchange control plane and create options for content reach. Six declared facilities provide more possible meeting points than one office or one cage. The edge is not a single anonymous uplink.

But three qualifications determine how much resilience this creates.

First, the data are entity-maintained. PeeringDB records what the network or authorised exchange sources declare; it is not a continuous audit of physical circuits. “Operational” means the connection is represented as in service. It does not establish current utilisation, service-level terms, last-mile carrier, route count or failover testing.

Second, an exchange port is not full transit. A route server can provide reach to participating networks that choose to peer. It does not necessarily provide every route on the internet. Om Shiv Sai still needs provider connectivity for destinations not reached through settlement-free or bilateral peering. RIPEstat's ASN-neighbours endpoint found six neighbours on 10 July 2026, but classified only AS137085 on the provider side and marked five relationships uncertain. The topology supports a visible upstream dependency, not a verified two-provider design.

Third, metro diversity can collapse into one shared path before traffic reaches the customer. The DE-CIX and Extreme IX ports may be in different exchange fabrics, yet leased transport could share a duct, a carrier backbone, a meet-me room, a local power feed or one aggregation router. Two facility names can even refer to operationally related sites within the same metropolitan risk zone. Without route drawings and carrier circuit IDs, the correct phrase is “two declared exchange connections,” not “fully redundant upstream routes.”

This distinction is central to the bill. Peering can reduce average upstream cost and latency, making a low retail price viable. It can also reduce load on paid transit. During a failure, however, all traffic that used the lost port must fit through surviving paths. A 30 Gbps port plus a 10 Gbps port is 40 Gbps of installed interface rate, not 40 Gbps of guaranteed usable capacity. If the larger port fails at peak, the smaller port may be physically unable to absorb the load even if routing converges perfectly.

The company should therefore be judged on surviving capacity, not summed labels. Useful disclosure would show peak utilisation on each exchange and transit link, the largest single failure, traffic shifted during that failure, and the headroom remaining after convergence. Nothing public reviewed here supplies those values. The network edge earns a Medium evidence grade because it is active, dual-stack, authorised and multiply connected at exchanges. Physical and capacity resilience remain Weak.

The failure ladder begins at the wall socket and ends with a route table

A customer-visible outage can originate at several layers. Keeping them separate makes recovery responsibilities clearer.

Customer equipment and power. The first test is whether the router, optical terminal, cable modem or radio is powered and synchronised. If only one customer is affected, a failed adapter, damaged drop or misconfigured device is more likely than a global route loss. The app can support account checks, but a field visit may still be required. Public terms do not say whether Om Shiv Sai supplies a battery, stocks replacement devices locally or promises same-day replacement.

Building or street access. If several nearby accounts fail together, the shared building switch, distribution point, splitter, radio sector or feeder becomes a likely fault domain. Repair may require landlord access, a lift, traffic control, pole permission or a cable crew. The provider's licence and ASN do not confer automatic access to every support structure. A clear operator boundary would identify who owns the customer drop, who owns the feeder and who has authority to open the enclosure.

Powered aggregation. A neighbourhood switch or optical line terminal can fail even when the fibre is intact. Grid power, batteries, rectifiers and cooling then become connectivity assets. A backup system has four separate properties: rated capacity, actual battery health, fuel or recharge logistics, and the load attached to it. A website phrase such as “redundant systems” proves none of them. The relevant operational number is runtime under the real node load, followed by tested start and recharge performance.

Aggregation congestion. A network can remain technically reachable while becoming unusable. Lost links redirect traffic, and surviving interfaces fill. Video buffering, packet loss and high latency may appear before a complete outage. The retail plan rate is not the diagnostic. Engineers need customer-serving-node utilisation, packet loss, queue depth and the location of the bottleneck. No Om Shiv Sai-specific QoS row or public status history supplies this detail.

Metro transport loss. If the access network is healthy but all external destinations fail, the transport path to Mumbai interconnection may be cut or de-energised. A second exchange port helps only if the route to it is independent and has enough headroom. Recovery may belong to a leased-line carrier rather than Om Shiv Sai's field team. Customers still call Om Shiv Sai because it owns the retail service, so escalation quality becomes part of the product even when another company performs the physical repair.

Transit or BGP failure. Routes can be withdrawn, filtered or propagated incorrectly while local links remain lit. Valid RPKI reduces one class of origin error but does not prevent every leak, bad path or upstream outage. A provider-side adjacency to AS137085 was visible in current route data. Additional route-server and peer adjacencies may preserve access to some destinations while full-transit loss isolates others. The symptom can therefore be selective: local caches or peered content work while less-connected destinations fail.

Exchange or facility failure. A port, switch fabric, cross-connect, router, meet-me room or facility power system can fail. Presence at more than one exchange and facility gives the operator options, but the actual response depends on where routers sit and how circuits are built. Six facility records do not prove six active edge routers. The public data do not disclose chassis count, line-card separation, cross-connect diversity or remote-hands agreements.

Compound failure. The hardest incidents cross layers. A power event can disable a local node while also affecting a mobile network that technicians use for coordination. Heavy rain can damage an aerial route while slowing access to the fault. A fibre cut can push traffic onto a smaller backup path, creating congestion that looks like a second fault. A large incident can consume the spare cable, optical modules and technician hours intended for ordinary repair.

Who is affected depends on where the ladder breaks. A customer-device fault affects one account. A building node can affect dozens. A feeder or aggregation failure can affect a neighbourhood. A shared metro tail can affect most of the footprint. A route-origin or full-transit failure can affect every subscriber's reachability even when local Wi-Fi appears normal. This scaling is why physical maps and fault-domain counts matter more than broad claims of reliability.

Field repair is not a support add-on; it is capacity

The OSS Internet app's complaint function shows that the provider has a formal channel for faults. Google Play reported more than 1,000 downloads when reviewed, and the listing said customers can register and track complaints, renew service, inspect usage and download receipts. Those functions reduce some support cost and give customers a record. They do not repair a cable.

For a regional operator, the field organisation is a form of capacity just as surely as an exchange port. One technician can complete only a finite number of jobs per shift. Travel time, building access, weather, splicing complexity and spare availability determine the queue. Growth from 900 to roughly 4,000 reported subscribers multiplies the installed base that can fail. If crew size and spares do not scale with that base, ordinary fault performance may appear acceptable until several failures arrive together.

The most important undisclosed number is crew concurrency: how many separate infrastructure faults can be investigated and repaired at the same time. A 24-hour phone line is not the same as a 24-hour field shift. An app ticket acknowledged at midnight is not the same as a technician dispatched. A contractor roster is not the same as stocked vehicles and access rights. Recovery evidence should distinguish response, arrival, temporary restoration and permanent repair.

Spare inventory also shapes restoration. Customer devices, power supplies, optical modules, switches, fibre closures, connectors and drop cable fail in different ways. A replacement device can restore one account in minutes if it is nearby and preconfigured. A feeder repair may require locating the fault, obtaining permission, splicing many fibres and testing every branch. A small operator can sensibly hold limited expensive spares, but then supplier and courier lead times become part of outage risk.

No public source reviewed here states mean time to repair, next-working-day performance, night coverage, depot location, crew count or spare levels. The company's terms say availability is not guaranteed and maintenance may temporarily interrupt service. That is ordinary contractual language, not a performance record. The correct conclusion is that local support is visibly part of the service but its recovery capacity is unverified.

Customers and enterprise buyers can ask practical questions without requiring confidential architecture. Is there a published support window? Are multi-customer incidents acknowledged on a status page? Is there a target for technician arrival? Does the operator distinguish customer-device faults from area outages? Is service credit available? Are backup devices held in the local service area? These answers would turn a support promise into measurable operating evidence.

The website is evidence of a sales surface, not evidence of resilience

Om Shiv Sai's public website deserves an explicit evidence downgrade because it mixes current company details with obvious template remnants. Its home page claims both more than 30 years and, elsewhere, 20 years of service, although the corporate and licensing evidence dates the current company and ISP authorisation much later. It labels a coverage section “San Diego,” names “MaxiCom” in testimonials, contains Latin filler text and refers to a “Lanet” connection. Those statements cannot safely describe Om Shiv Sai's network.

This does not make every company-provided fact false. The contact page's legal name, licence number, CIN and Maharashtra address align with external records. The plans page is specific, priced in rupees and consistent with a customer offer. The app links back to the same domain and gives matching developer details. Corroborated identity and tariff facts remain usable.

What must be rejected are uncorroborated engineering superlatives. The site says its systems are redundant and support is available around the clock. It does not identify two upstream carriers, a ring, a backup route, a battery specification, a network operations centre schedule or an outage history. Because the same pages retain unrelated template copy, these claims carry less weight than they would on a maintained engineering or regulatory disclosure.

The contaminated site also matters operationally. A prospective customer cannot tell which coverage statements apply locally, whether the plan table is current, or which reliability language is contractual. Clear public information is part of outage handling: during a fault, customers need a current contact, status notice and expected restoration time. The app is a better signal than the generic web prose, but neither provides a public incident archive.

The reasonable interpretation is neither “the website proves resilience” nor “the website proves the network is poor.” It proves that the operator has a customer-facing domain whose content controls need attention. Current routes, licence records and subscriber reports independently establish operation. Physical resilience remains unanswered.

What would raise the evidence grade

The fastest route from Weak to Medium physical-resilience evidence is not another marketing claim. It is a compact set of verifiable operational disclosures.

First, publish the service boundary. A locality list or address checker should distinguish areas directly served under Om Shiv Sai's access plant from areas reached through local operators or wholesale infrastructure. It should state the access medium available at an address without implying that every plan is technically available everywhere.

Second, disclose fault domains at a useful level. The operator need not reveal security-sensitive cable coordinates. It can state how many aggregation zones exist, whether critical zones have ring protection, whether exchange paths use separate carriers and entrances, and which elements remain single points of failure. A map of shared-risk groups is more valuable than a decorative network map.

Third, state surviving capacity. For each major edge failure, the company could report the largest expected traffic shift and whether the remaining links can carry peak demand within a chosen utilisation threshold. This would convert the 10 Gbps and 30 Gbps port declarations from installed rates into recovery evidence.

Fourth, show upstream diversity. Route collectors currently make AS137085 the clearest provider-side dependency. If a second full-transit provider exists, a public routing policy, looking glass, current route data or provider letter could establish it. Peering connections should be described separately because they may not replace full transit.

Fifth, report power and field readiness. Useful measures include backup runtime at critical nodes, battery-test frequency, generator or portable-power coverage, technician concurrency, spare-device stock, mean restoration time and the longest area outage in the reporting period. These measures can be aggregated; they do not require exposing employee identities or exact site locations.

Sixth, maintain a status history. The July 2026 route gaps are a signal that cannot be interpreted from route data alone. A public incident record could say whether customers were affected, which layer failed and when full service returned. Over time, this would let buyers compare architecture claims with actual recovery.

Until those disclosures exist, the best monitoring set is modest: current DoT authorisation, TRAI subscriber counts, APNIC resource changes, RPKI validity, route visibility, provider-side neighbours, exchange-port status, app maintenance and credible customer notices. Unofficial topology services can flag changes, but no single commercial label should be promoted into a verified contract or physical route.

A functioning regional edge rests on an undisclosed local machine

Om Shiv Sai is not a paper ISP. Official records support a current Category C authorisation; TRAI supports a subscriber base that expanded beyond its long-reported 900 accounts; APNIC supports an active dual-stack ASN; route collectors support current global visibility; and exchange records support two operational Mumbai peering ports. The company-region-global-type-regional-isp category and all three controlled topics are therefore evidence-supported.

The evidence does not support the global region label, the shortened legal name, universal fibre access or a claim of proven physical redundancy. The local network remains largely invisible: no trustworthy coverage map, asset inventory, route separation, upstream contract list, power design, crew roster, spare policy or restoration record is public. Two exchanges and valid routes make the internet edge credible. They do not remove the dependence on the first shared cable, powered node or technician between that edge and a customer.

That is the practical meaning of the title. A monthly connectivity bill buys access to AS149214, but it also buys the operator's ability to maintain a neighbourhood path, keep aggregation powered, preserve enough surviving upstream capacity and reach a fault with the right spare. The first half of that chain is visible. The recovery half is not. On the public evidence available in July 2026, current operation merits a Medium grade and physical resilience a Weak grade.