Summary

  • RekorNet's own pages publish a 99.9% uptime figure, 7/24 technical-support language and a statement that faults requiring on-site intervention are resolved within 48 hours. These are three different operational commitments, not one end-to-end SLA.
  • The company markets wireless, VDSL/fibre, building-fibre and Metro Ethernet services. A buyer should therefore attach measurement, escalation, dispatch, restoration and remedy terms to the exact access mode and failure class being purchased.

Three public commitments need separate definitions

RekorNet's About page describes the company as a Balikesir-based internet service provider authorised by Türkiye's Information and Communication Technologies Authority. It lists wireless access, VDSL, fibre, Metro Ethernet and building-oriented fibre services. The same page displays a 99.9% uptime figure and says technical support is available 7/24.

The company's FAQ adds a more specific operational statement. When a fault requires intervention at the customer's premises, it says the fault is resolved within 48 hours; system-level problems can receive immediate intervention around the clock. RekorNet's contact page, meanwhile, publishes weekday hours of 09:00–18:00 and Saturday hours of 09:00–16:00, alongside call-centre numbers and a support email address.

These statements can all be true while describing different queues. A support mailbox can accept a case at any hour without guaranteeing that a field team will arrive at any hour. Remote intervention can begin immediately without establishing when service will be restored. A 48-hour field statement can govern a customer-premises fault without defining the measurement period or exclusions behind a 99.9% availability figure.

The public pages do not, by themselves, establish how those clocks interact for a particular contract. That is the first due-diligence gap: not whether RekorNet publishes service language, but which failure starts which clock, who owns the hand-off and what remedy follows if the clock expires.

Access mode decides where the fault can live

RekorNet's consumer offer page separates packages marketed as infrastructure-free wireless access from VDSL and fibre offers. It also publishes installation and transfer charges, modem rental and equipment-return obligations. Its corporate page advertises business connectivity, static IP and SLA language across a range of speed profiles.

That range matters more than the common logo. A fixed-wireless customer depends on radio placement, alignment, spectrum conditions, power and the operator's aggregation path. A VDSL or wholesale fibre service may add a third-party access network and its repair process. Building fibre introduces common equipment and in-building distribution. Metro Ethernet has another demarcation and may carry separately negotiated service terms.

The result is not that one access mode is inherently reliable and another is not. It is that an uptime percentage without the product boundary is hard to use. A buyer needs to know whether the calculation begins at the customer-facing port, excludes customer equipment, includes the wholesale tail, counts planned maintenance, treats partial degradation as downtime and measures each circuit or an aggregate platform.

The public corporate table itself rewards careful reading. It calls the offer family symmetric, while one listed package shows 200 Mbps downstream and 100 Mbps upstream. That does not prove a delivery defect. It does show why the quotation, order form and technical schedule—not a section heading—must define the service purchased.

Registry records identify an observation surface

The RIPE record for AS204348 uses the name REKORNET, links the autonomous system to ORG-RTVB4-RIPE and records ASSIGNED status. Its published policy includes import statements from AS15924, AS9121 and AS208972, with export statements announcing AS204348. The corresponding RIPE organisation object names Rekornet Telekomunikasyon Ve Bilisim Hizmetleri Ltd. Sirketi, classifies it as an LIR and records the Balikesir address and country code TR.

RekorNet's operator-maintained PeeringDB entry also identifies AS204348 and lists an operational 10 Gbps connection at GIBIRIX with IPv4 and IPv6 addresses. These records make the network easier to investigate. They give an analyst an ASN, declared relationships and an exchange point at which to form observation questions.

They do not measure the purchased circuit. RIPE policy is not a live BGP feed, and PeeringDB is a self-maintained industry record rather than proof of present path preference, usable capacity or physical diversity. Neither record demonstrates that a customer's wireless or wholesale tail remains available when AS204348 is reachable elsewhere.

That boundary prevents a common category error. Network-resource evidence can identify the operator's public routing surface, but an access SLA must still say which segment is covered and how a customer proves an incident.

A buyer needs five separate receipts

The useful response is not to reject the public claims. It is to turn each into a verifiable receipt.

First, the service receipt should name the exact product, access medium, hand-off, committed rates and customer-premises equipment boundary. Second, the availability receipt should define the measurement point, interval, exclusions and treatment of degradation. Third, the support receipt should distinguish acknowledgement from diagnosis, remote action and escalation to a named duty function. Fourth, the field receipt should define dispatch and restoration clocks, coverage hours, access requirements and responsibility when a wholesale provider or building system is involved.

Fifth, the remedy receipt should specify credits, termination rights or other consequences when the contracted target is missed.

Those receipts also reveal economic trade-offs. No-commitment consumer plans and transparent equipment charges can reduce lock-in, while a business-grade restoration obligation costs money to staff, stock and enforce. A local provider can offer valuable proximity, but locality becomes an operational advantage only when the escalation and field model is explicit.

RekorNet's published material supplies useful starting points: named services, prices or quotation paths, support channels, a field-repair statement, an uptime headline and a visible network identity. The remaining work belongs in observation and contract. The 99.9% figure should be tested against the purchased service; it should never be expanded into a measured, network-wide result that the public sources do not provide.