From Fragmented Logs to a Trusted Data Foundation: The .ZA Registry Story
A registry with 600+ million rows of fragmented, decades-old data needed one trusted source of truth. We built it, and it now runs regulatory reporting, abuse detection and forecasting for the .ZA and .Africa namespaces.
The problem
Every registry is a data company whether it wants to be or not. Trust, transparency and accountability all rest on the quality of the underlying data. DNS Africa's reality a few years back looked nothing like that. Data was scattered across legacy systems, duplicated across sources, and stitched together through manual reports. Nobody could point to one number and say with confidence "that's correct." Regulatory reporting was slow and error-prone, and every new question about the business meant another manual pull.
Context
This started as a University of Pretoria final-year project: could a small team of CS students build a prototype that brought structure to the registry's data? The proof of concept surfaced both the opportunity and the scale of the mess. That project became the seed of what turned into Tyto's long-term analytics partnership with DNS Africa.
What Tyto built
A star-schema data warehouse as the single trusted source, holding domain history from 1994 to today, roughly 600 to 800 million rows. Built around one golden rule: once data is cleaned and verified, the numbers never change again.
On top of that warehouse: APIs and role-based dashboards for registrars, regulators and internal teams. Live streaming of DAMS/EPP commands and DNS request traffic. Financial data and forecasting for registrars. IP stats and web classification across every domain the registry holds (co.za, org.za, net.za, web.za, .africa and all RyCE domains). Updates run every 15 minutes, and the whole platform sits inside the registry's own environment, not in the cloud, so data sovereignty is never a question.
The pipeline follows the same discipline everywhere at Tyto: capture, clean, transform, verify. Clean before you engineer.
What this proves
That the "clean before you engineer" approach scales to genuinely large, messy, decades-old datasets, and holds up under regulatory scrutiny. What began as an automation project for compliance reporting turned into the operational backbone the registry now runs on daily, from abuse monitoring to renewal forecasting to a public-facing brand protection product.
- Regulatory reporting time cut from hours to minutes
- ~600–800 million rows spanning 1994 to present, queryable in milliseconds
- Real-time DNS abuse monitoring built on the same warehouse
- Predictive analytics for renewals and anomaly detection
- Platform now embedded as a standard layer across all DNS Africa deployments
- Spun out Brand Watch, a proactive phishing/look-alike domain monitoring product
Your business has a story like this waiting
It usually starts the same way — better analytics requested, a data foundation problem found.