Summary

  • NIC.br records tie Servicos de Infraestrutura e Datacenter to CNPJ 35.651.317/0001-72, AS272547, a directly allocated IPv4 /22 and an IPv6 /32. That is meaningful evidence of a real network role.
  • A dated routing view showed one upstream, Tyna Host's AS270353, while reverse-DNS administration also pointed into Tyna Host's orbit. The connection is operationally relevant, but it does not prove that the two organisations, contracts or facilities are interchangeable.
  • No public evidence pack connected the subject itself to a defined cloud product, customer portal, facility specification, measured uptime, recovery result or support performance. Procurement should price that uncertainty rather than treating the words in the company name as assurance.

Start with the resource holder, not the label

"Servicos de Infraestrutura e Datacenter" sounds like a product description translated into a company name. That creates an unusual diligence problem. A buyer may assume that it denotes a staffed facility, racks, power systems, virtualisation, storage and an operating team. The public record supports a narrower conclusion.

NIC.br's registration service identifies the organisation under CNPJ 35.651.317/0001-72. It records AS272547 as a direct allocation in Brazil, registered in December 2021, and links the organisation to 190.102.40.0/22 and 2804:84f0::/32. Separate records mark both address blocks active and associate them with the same autonomous system and company identifier. LACNIC's 2024 and 2026 electoral registers also list the company name among Brazilian member organisations.

This is not merely a directory echo. An autonomous system and directly allocated addresses create a public operating surface: routes can be observed, abuse and technical contacts can be resolved, and changes can be compared over time. The evidence establishes that the subject holds internet number resources and has appeared in the regional registry community. It does not establish what is sold to customers.

An ASN can belong to a hosting operator, a content network, an enterprise, an access provider or an organisation using a mixture of its own and third-party infrastructure. An IPv4 allocation says who is accountable for the registry record; it does not say who owns a server, where that server is installed, which customer occupies it or whether an application will recover after a failed host. The correct starting statement is therefore precise: Servicos de Infraestrutura e Datacenter is a verifiable Brazilian resource holder and route origin. "Datacenter operator" remains a claim to test against a specific service.

Routing evidence reveals both activity and concentration

The network was active at the time of review. BGP.tools described AS272547 as originating 19 IPv4 prefixes and one IPv6 prefix, with one upstream carrier: Tyna Host's AS270353. It also characterised the network as small. IPinfo independently described it as a single-homed autonomous system. Another route view, IPIP, showed 16 IPv4 prefixes and one IPv6 prefix during collection.

That variation is not a reason to choose the largest or smallest number. Route collectors observe different vantage points and can update at different times; more-specific announcements also make a route count different from a count of unique address blocks. The discrepancy is itself useful. Any capacity or resilience claim should be attached to a dated export, a measurement point and an explanation of which prefixes carry the customer's service.

The route table also reaches beyond the company's directly allocated /22. Public descriptions on observed prefixes included Tyna Host, Internet Utilities NA, Private Customer, Diamond IP Brokers and Brander Group. Those labels may describe customers, lessors, address holders or other commercial arrangements. They should not be converted into corporate ownership, hosted-customer lists or physical-location claims. They do show that AS272547's operating role involves address space with several public descriptions, making route authorisation and customer-boundary records important procurement evidence.

The single visible upstream matters more immediately. A sole provider relationship is not automatically a design failure: the upstream may itself have extensive path diversity, and a small network may deliberately buy a managed transit product. But there is no independent internet path if the customer service depends only on that one upstream relationship. A buyer should ask whether every relevant prefix was single-homed, whether a second path existed but was not visible, what capacity was committed, how denial-of-service traffic was handled and who had authority to change routing during an incident.

Routing visibility proves reachability at one layer. It does not prove storage durability, hypervisor health, DNS availability, cooling, generator autonomy or a working restore. The ASN strengthens the case for further diligence because there is something concrete to inspect. It cannot carry the whole availability argument.

The Tyna Host connection must be explained, not assumed

The most consequential public relationship is with Tyna Host. It appears in three distinct places. BGP.tools observed AS270353 as AS272547's only upstream. NIC.br's record for the directly allocated IPv4 block showed reverse-DNS delegations using names under svrbrasil.com.br. The registration record for that domain names a Tyna Host administrative contact and a different registrant from the company that holds AS272547.

These facts support an operational connection. They do not establish an acquisition, common ownership, agency or a right to apply one party's promises to the other. Even the infrastructure labels need care. Tyna Host's public site advertises Brazilian VPS, dedicated servers, colocation, denial-of-service protection, round-the-clock support and an initial ticket response in under 15 minutes. Those are Tyna Host claims. The collected material did not show that CNPJ 35.651.317/0001-72 is the contracting entity for those offers or that the promises govern a service sold under the Servicos de Infraestrutura e Datacenter name.

There is also a small but telling control detail. At collection time, the svrbrasil.com.br host answered over plain HTTP, while a normal HTTPS request failed hostname validation because the certificate did not match the requested domain. That is not evidence of a customer outage; the domain may exist mainly for nameserver functions rather than as a public storefront. It does show why operational artefacts should be interpreted according to their actual role. A working reverse-DNS delegation, a brochure site and a customer control plane are three different surfaces.

Before procurement, the parties should draw the relationship in one page. Who issues the invoice? Who supplies the rack, power, transit, address space, mitigation and remote hands? Which organisation receives a security notice? Whose staff can access a server? Which support desk owns restoration rather than merely forwarding the case? If the subject resells or layers services on Tyna Host infrastructure, that may be commercially reasonable. The value lies in a clear responsibility chain, not in pretending the dependency does not exist.

A service name is not a service boundary

The public evidence did not identify a subject-owned catalogue, account portal, facility specification, service agreement or status history. That leaves several materially different products hiding behind one broad name.

A buyer of virtual private servers depends on the hypervisor, storage, network policy, host capacity and provisioning system. A dedicated-server customer depends on hardware replacement, spare parts, remote access and rebuild procedures. A colocation customer retains control of the machine but transfers power, cooling, physical security, cross-connects and hands-on intervention to the facility operator. A managed-infrastructure customer may transfer operating-system patching, backups and monitoring as well. Each product allocates labour and failure differently.

The distinction is essential when comparing price. A hyperscale cloud offers deep automation, multiple regions and extensive control APIs, but can create architecture, egress and specialist-labour costs. Colocation can give a customer more hardware control and predictable capacity while leaving it responsible for equipment, spares and lifecycle management. Self-run infrastructure maximises freedom but makes the customer staff power, networking, security, monitoring and incident response. A regional provider can win by combining local placement, human support and a simpler bill.

It earns that premium only when the service boundary removes more supervisory work than the dependency adds.

Servicos de Infraestrutura e Datacenter cannot be placed fairly in that comparison until the product is named. Procurement should request a bill of service that separates compute, storage, network, backup, mitigation, software licensing, monitoring, remote hands and support. It should identify dedicated and shared components, capacity limits, usage charges and termination costs. A low headline price is uninformative if restore work, additional addresses, traffic bursts or after-hours intervention sit outside it.

Automation moves labour into the control plane

Cloud and hosting services replace manual tasks only when customers can see and govern the state that automation creates. Provisioning a virtual machine, assigning an address, changing a firewall rule, scheduling a backup and raising a support case may all happen through separate systems. The productivity gain depends on whether those systems agree about the same asset.

A credible control plane should preserve who requested a change, what was changed, when it completed, what it cost and how it can be reversed. It should support distinct administrator and auditor roles, strong authentication, credential revocation, activity export and a usable path when the portal itself is unavailable. Billing records should reconcile with live capacity. Backup jobs should identify the protected source and recovery point. A ticket should carry the affected server, route or storage volume rather than begin with a search for basic account context.

None of this was demonstrated for the subject in public. That absence does not prove the controls are missing; it defines the demonstration a buyer needs. Ask the provider to provision a resource, change access, export the event history, recover from a failed action, restore data to a clean target and reconcile the result with the invoice. Then repeat the exercise through the emergency support path. A product tour shows the happy path. A controlled failure shows whether automation has reduced labour or merely moved it behind an opaque screen.

The upstream relationship should be part of the same test. If a route is withdrawn, does the customer portal show a network incident or only an unreachable server? Can support connect the resource identifier to the affected prefix? Who contacts AS270353, and what update interval applies while the upstream investigates? Operating assurance appears when technical state and human authority meet in one workflow.

Brazilian registration does not settle data locality

The subject's registry country is Brazil, and its directly allocated addresses are Brazilian internet resources. Tyna Host markets infrastructure in Brazil. Those facts may make local hosting plausible, but they do not locate customer data.

The physical server, backup copy, snapshot, monitoring platform, support attachment and administrative session can each be somewhere different. Additional routes originated by AS272547 carried public descriptions associated with other organisations and countries. A route description is not reliable server geolocation, just as a Brazilian ASN is not proof that every workload stays in Brazil. Locality must be established at the service and data-class level.

A buyer with sovereignty requirements should obtain a location map for production data, replicas, backups, logs, ticket content and control services. The map should name the legal entity responsible at each point, any subcontractor, the access geography, encryption and key ownership, retention, deletion and the circumstances under which failover can move data. It should also explain whether address space is provider-owned, leased or supplied by another party and what happens to those addresses on exit.

This is not paperwork around the technical service; it is part of recoverability. A backup held under a separate account and failure domain can improve resilience. A backup that shares credentials, operators, power or storage with production can preserve the same failure. The useful proof is a successful, dated restore against an agreed recovery point and recovery time, followed by evidence that the restored data stayed within the approved boundary.

Support needs authority, targets and retained evidence

The visible Tyna Host site offers a support channel and a rapid initial-response claim. It would be a mistake to attach that target automatically to Servicos de Infraestrutura e Datacenter. The relevant promise is the one made by the contracting entity for the purchased service, with severity, clock and remedy defined.

An acknowledgement target answers only the first question: did someone receive the case? Production support also needs a restoration objective, escalation ladder, communication interval and authority to act across the facility, virtualisation layer and upstream network. If the first desk cannot alter a route, replace a disk or dispatch remote hands, the contract should state how quickly the case reaches someone who can.

Buyers should request twelve months of service-specific availability and incident data, measured from an agreed observation point. They should inspect a sample incident account for timestamps, affected assets, decisions, customer communications and corrective actions. They should test the telephone or ticket route before launch and verify that after-hours staff can identify the customer, the service and the escalation owner. Support has value when it reduces diagnosis and decision time, not merely when a queue is open.

The same evidence protects both sides. Asset identifiers, route snapshots, change logs and ticket history can distinguish a provider outage from a customer configuration error. A responsibility matrix can show whether the provider owns the hypervisor while the customer owns the operating system. Clear records reduce argument when minutes matter.

Buy the proof that connects the layers

Servicos de Infraestrutura e Datacenter has a stronger public foundation than its generic name initially suggests. It is tied to a Brazilian company identifier, an active ASN, directly allocated IPv4 and IPv6 resources, current route origin and repeated LACNIC membership records. The Tyna Host relationship is also visible enough to become a concrete diligence subject rather than an unnamed dependency.

The remaining gap is the connection between those facts and a customer outcome. A defensible evidence pack should identify the contracting and operating parties; map the bought service to facility, hardware, network and software; provide dated routing and capacity information; define every data location; show backup and recovery tests; and set response, restoration, escalation and remedy terms. It should also preserve an exit route for data, configurations and any addresses the customer does not own.

The decision is not whether AS272547 exists. It does. The decision is whether the company can connect that network identity to the particular workload, support team and recovery obligation for which a customer will pay. Until that connection is documented and tested, the name describes an ambition. The resource trail earns a serious assessment; the contract and operating evidence must earn the trust.