Dataprev - social security as a computing problem

Dataprev was created by Law 6.125 of 4 November 1974, as a public company under the social security ministry, headquartered in Rio de Janeiro. It was not built from nothing: it was assembled from the data processing centres of the separate social security institutes that had existed until 1964, which makes its founding an act of consolidation rather than of creation - the state had the computing already and decided it should be one organisation.

The inherited hardware came with a contract. In 1975 the agreement between the social security institute and Burroughs passed to Dataprev, which took over running the installed machines in Rio de Janeiro and Sao Paulo. That is a detail worth keeping for anyone who works on public sector estates: the company began by operating somebody else's procurement decision, which is how most public computing actually starts.

What it does now is the part that changes how you read it. It holds the Cadastro Nacional de Informacoes Sociais, the largest social database in the country, and it processes the monthly payment of every retiree and beneficiary of the national social security system - by its own account the largest payroll in the world, never once late in more than forty years. In roughly five hundred Brazilian municipalities those benefits amount to more than a quarter of local economic activity.

That last figure is what makes this an infrastructure entry rather than a government one. An outage at Dataprev is not an information technology incident, it is a regional economic event in hundreds of towns simultaneously, which is the clearest illustration available of why availability sits alongside confidentiality and integrity rather than beneath them. The company also ran the emergency benefit payments of 2020 at national scale on short notice, and operates three data centres certified to TIER III.