Payments are experiencing issues due to temporary restrictions in Russia. If your payment does not go through, please submit a support request.Our support team is available 24/7 — we are always here to help with hosting and server issues.We are now accepting requests for dedicated server rental and colocation services in our data center.Reminder: we recommend enabling backups for additional data protection.A new VPS/VDS lineup with NVMe storage and improved performance is now available.Maintenance work on some servers has been completed. All services are operating normally.
Article5 min readViews0

Two DNS Servers: Why Different Names Don't Mean Independence

Two NS records in domain settings may depend on a single platform, network, or control panel. We examine which failure points to check and how to accept a DNS scheme without experimenting on a live site.

Comments 0

Mikhail compares two independent routes to DNS servers on a network diagram
In this article

The registrar's panel lists two DNS servers: ns1.example and ns2.example. Formally, the requirement is met. However, both names may point to nodes in the same data center, rely on a single network uplink, and be managed through one control panel. If the shared component fails, both will disappear simultaneously. Therefore, the number of NS records does not indicate fault tolerance.

It is more useful to ask a different question: what single failure would leave the domain without an authoritative answer? Below is a verification scheme for website owners and the team accepting DNS from a hosting provider or contractor. It does not promise absolute availability: boundaries depend on the service architecture, routing, monitoring, and change management procedures.

What exactly must survive a failure

An authoritative DNS server is responsible for its zone's data. A recursive resolver can query any available server listed for the domain. The terms "primary" and "secondary" describe the origin of the zone data: the secondary receives it via zone transfer. For an external query, both can be equal sources of the answer.

This leads to two distinct requirements. First, servers must remain reachable even if one site or network path fails. Second, they must hold a consistent, up-to-date zone. Spreading addresses but forgetting to update a single node is also problematic: some visitors will receive stale records.

Two names can point to a single point of failure

Different names and even different IP addresses are only the beginning of the check. Nodes may reside in the same rack, draw power from the same line, route through the same operator, or depend on a single management account. DNS operational guidelines explicitly require accounting for physical and topological separation: two servers in the same room or on the same local network do not provide full redundancy.

Diagram: two DNS names and different IPs depend on a single site, network junction, or control panel
Different records do not eliminate a shared infrastructure dependency.

Drawing conclusions from a single indicator is premature. A matching autonomous system or a single public address may warrant a closer look at the architecture, but they do not prove how it is built: for example, a distributed anycast service intentionally advertises a single address from multiple points. Conversely, two different addresses do not rule out shared power, panels, or configuration sources. Therefore, relying on the external appearance of records should be replaced with a description of dependencies and verifiable conditions.

Separate dependencies, not just names

For a standard website, I would divide the check into five layers: compute node, site and power, network and routing, management system, and zone data source. If two NS instances are independent only at the first layer, a failure of the shared control panel or an erroneous zone publication will still affect both.

  • Nodes are placed so that a failure of one site does not disable all responses.
  • Network paths do not converge at a single mandatory point outside the team's control.
  • Zone changes reach every authoritative server, and the SOA serial number matches.
  • Access to the control panel and disaster recovery do not depend on a single account without a backup owner.
  • Monitoring checks authoritative responses from independent networks, not just site availability from the office.

Different DNS providers can reduce overall infrastructure dependency, but they add operational complexity: automation formats, update delays, access rights, and DNSSEC must be aligned. A single operator with a documented distributed architecture can also be a reasonable option. The choice depends not on the number of logos, but on which failures your specific scheme must survive and who is responsible for its updates.

How to accept a scheme without an outage on the live domain

There is no need to disable production DNS servers for verification. First, gather facts and reproduce the change on a test zone where an error will not disrupt the store. For acceptance testing, it is sufficient to define the expected result in advance and check each authoritative server individually.

Acceptance scheme for independent DNS: two servers on different platforms receive the same zone and are checked from independent networks
Verification separates zone consistency from network path availability.
  • Compare the delegation in the parent zone with the list of servers supported by the team. An extra old NS is a risk just as much as a missing new one.
  • Record addresses, platforms, networks, the DNS operator, managing accounts, and the zone source. An unknown dependency must remain an open risk, not an assumption of fault tolerance.
  • Change a harmless record in the test zone and wait for the expected SOA serial number on each server. Also verify that the answers match in meaning, not just that the nodes respond.
  • Request verification from two independent networks or monitoring points. Latency alone is not the main criterion: a correct authoritative response and the absence of a single unavailable path are more important.
  • Describe the rollback to the previous configuration, access owners, and the communication channel with the provider. If recovery relies on a single person or a mailbox within the same domain, this is a separate point of failure.

A separate plan is required for DNSSEC and provider migration: the sequence for publishing keys and records in the parent zone is as critical as server availability. Such a transition should not be combined with the initial failover test. First, validate the current configuration, then change one layer at a time.

A criterion that fits into a single question

Two DNS servers can be considered a meaningful backup only after answering this question: which single point of failure would eliminate all authoritative responses, and who verified this? If the answer is unknown, different names only confirm delegation settings, not independence. A documented architecture, a synchronized zone, and monitoring from independent points provide stronger grounds for trusting the scheme—without guaranteeing it will survive every scenario.

Discussion 0

Share your experience and ask questions. Comments without links appear after editorial review.

No comments yet. Start the discussion.