A registry launch rarely fails because of policy ambition. It fails when the operating model and the platform underneath it do not match. That is why the back-end question deserves attention early, before delegation, migration planning or registrar onboarding is already underway.
Operators usually frame the choice as three options: join a shared registry system, buy a managed service, or build and host a platform themselves. Each one carries different consequences for time to market, compliance posture, integration flexibility, resilience and long-term dependency. This guide compares them side by side and offers a set of questions that will hold up in any procurement process.
The registry back-end decision: build, host or share?
A registry back-end is the system of record for every domain in a namespace. It holds registrations, contacts, hosts and DNSSEC data, enforces policy, connects registrars through EPP, publishes zone data and feeds reporting, billing and escrow. Whatever you choose will shape how your registry operates for years, because replacing it later means a migration with real risk to registrars and registrants.
That lasting impact is the reason to treat this as a strategic decision. A platform that looks cost-effective during procurement can become restrictive once transaction volumes rise, premium programmes expand or policy exceptions start to pile up. A platform that looks powerful on paper can turn into a staffing problem if your team cannot run it.
Three operating models compared: shared, managed and self-hosted
Shared registry system
A shared registry system is a multi-tenant environment where several registries run on the same core platform. Infrastructure, maintenance and product development are spread across many operators, so standard registry functions arrive faster and at lower upfront cost.
Managed registry service
In a managed model, a provider operates the platform on your behalf, often with dedicated configuration, monitoring, support and release management. You keep ownership of policy and commercial decisions while the provider carries day-to-day technical operations.
Self-hosted or custom platform
Self-hosting means running registry software on infrastructure you control, sometimes built or heavily tailored for one namespace. It offers the most control over data locality, release cycles and integrations, and it also asks the most of your organisation.
The core trade-off between the three is control against operational burden. Self-hosting demands round-the-clock monitoring, security patching, database administration, DNS and DNSSEC expertise, escrow processes, incident response and a team that can cover leave and turnover. If those skills are not already in place, they have to be hired, trained and retained.
Where a shared registry system makes sense
For many ccTLDs, emerging gTLD operators and new namespace launches, a shared registry system is a practical starting point. It reduces implementation friction and shortens the path from contract to production. If your priority is stable operation with established lifecycle functions such as EPP, WHOIS or RDAP, DNSSEC, reporting, billing support and registrar connectivity, a shared environment can cover them efficiently.
The model works best when policy rules are fairly standard. If your registration lifecycle, pricing logic and compliance controls follow established registry patterns, funding extensive custom development adds little. You get less engineering overhead, fewer moving parts and a roadmap managed by a specialist.
This is the space where rEngin sits. rEngin is the core of the DNS Business registry system, so operators who choose a shared or managed approach with us are running on the same foundation that supports our own registry services. Shared does not have to mean simplistic: a well-designed platform can support significant scale, registrar automation and strong security controls. The real test is how far it can bend before your business model starts asking for exceptions.
When a custom or self-hosted platform is the stronger option
A custom platform makes more sense when the namespace is unusual, tightly regulated or commercially differentiated. Some operators need specialised validation workflows, local policy enforcement, multi-step approval chains, complex premium inventory logic or integrations with national databases and identity systems. In those cases, working around product limitations can cost more than designing the right platform from the start.
Custom and self-hosted environments also appeal to operators that need control over release cycles, security architecture, data locality and platform evolution. If a board, ministry or enterprise governance team sets strict operational requirements, a generalised platform may involve too many compromises.
The price is greater investment and a more demanding implementation. Customisation increases design effort, testing scope and change management, and it can create dependency on a narrow support structure if the builder lacks real registry operating depth.
Cost is about capacity, not licensing
The most common procurement mistake is comparing options on first-year price alone. A shared platform usually lowers entry cost, which can be the right call for operators that need to conserve capital or prove market viability first.
Lower entry cost can still lead to long-term friction if the platform forces manual workarounds or delays product changes. When your policy team keeps asking for exceptions and your registrar channel needs tailored workflows, inefficiency starts to eat the savings.
Self-hosting moves cost from licence fees to people and infrastructure. Count the engineers, the on-call rota, hardware or cloud capacity across more than one site, security tooling, audits and the time senior staff spend on platform decisions. A useful question for any option is which one produces the least operational drag over five years, given the team you actually have.
Security, compliance, resilience and control
Registry infrastructure is never neutral from a compliance perspective. Jurisdictional requirements, ICANN obligations for gTLDs, local data handling rules, auditability, abuse controls and continuity expectations all shape the choice. Our overview of what makes registry software compliant covers the requirements in more depth.
Shared and managed systems can deliver strong compliance and resilience when the provider runs disciplined operational controls, because the investment in security practice, redundancy and response procedures has already been made. Custom platforms offer precision instead: if your obligations call for very specific controls, reporting formats or hosting arrangements, tailoring may be the only sensible route.
Either way, control without operational rigour becomes a liability. More authority over the platform means more responsibility for governance, documentation, testing and vendor accountability. For resilience specifically, see our guide to registry failover architecture [link: 239408].
Operator control and registrar experience
Two groups live with your platform choice every day: your own operations team and the registrars who sell your names. Both need tools that fit their work.
Operators need a clear view of the registry: who the registrars are, what state their accounts are in, which domains need attention and how registry functions are configured. In rEngin, that is the job of rEngin Console, where registry operators manage registrars, domains and registry functions.
Registrars and resellers need a dependable way to handle their side of the relationship without raising support tickets for routine tasks. rEngin Portal gives registrars and resellers a place to manage their account, onboarding and domain-related functions. Whatever platform you evaluate, ask to see both views, because a registry that is easy to run but hard to sell through will struggle to grow its channel. Our reseller solutions page explains how we approach the channel side.
rEngin Portal version 3 is our latest release, already available at ZACR, RyCE, Gateway SRS and Registry Africa. The next deployment will be for ZARC; ZARC and Registry Africa are the two largest registries in the DNS Business portfolio.
Questions to ask any provider, including commercial terms
Product demos tend to show the happy path. These questions expose how a platform and its operator behave when things are less tidy:
- Which registry functions are standard, and which require custom work or paid change requests?
- How are releases scheduled, tested and communicated to registrars?
- What monitoring, incident response and escalation processes are in place, and who staffs them?
- How are escrow, data protection and audit requirements handled?
- What does registrar onboarding look like, from credentials to production access?
- How is data exported if you leave, and in what format?
Commercial terms deserve the same scrutiny. Check how pricing changes as domain volumes grow, what service levels are contractually committed, how termination and exit assistance work, and who owns configuration and customisation. A low headline price with weak exit terms can cost more than a higher price with clear ones.
Migration risk and choosing for change
For established registries, this decision is usually tied to a migration. Legacy systems may limit automation, frustrate registrars or create security and reporting gaps, so platform selection and registry migration planning become one project.
Proven, standardised target environments tend to lower migration risk, since data mapping, registrar onboarding and cutover planning can follow mature playbooks. Custom migrations can succeed too, but more tailored logic means more edge cases and more testing. Operators running a country-code or generic namespace will find the specific considerations in our ccTLD vs gTLD operations guide [link: 239456].
Finally, choose for change. Your registry will not look the same in five years: policies shift, registrar channels evolve and new services appear. The right back-end still fits when your registry becomes more complex than it is today. Start with what cannot be compromised, such as policy enforcement, registrar experience, compliance, security and service levels, then test each model against that list instead of against a feature checklist. Our team at DNS Business has run registry back-ends through exactly these transitions, and we are happy to compare notes.
See rEngin for yourself
If you are weighing shared, managed or self-hosted options, a walkthrough of rEngin Console and rEngin Portal is the quickest way to see whether the model fits your registry. Book a rEngin demo.


