Arsitektur DataOps dan Aturan Data Wajib di Indonesia

The search for DataOps architecture almost always ends in box-and-arrow diagrams: data sources on the left, dashboards on the right, and several layers in between. While the diagram is not wrong, it does not answer the truly important question of what must be fulfilled by a layer before it can be considered complete. In Indonesia, part of the answer is already written in regulations, and that part is the least read by data teams.

DataOps places quality gates within the flow, not at the end

DataOps is a way to manage data flows like a production line: there is orchestration from upstream to downstream, there are versions for each component, and there are self-running checks. The DataOps Manifesto managed by DataKitchen contains 18 principles and five values. Four of these principles are the most decisive in shaping the architecture: end-to-end orchestration, reproducibility of results through versioning of all components, automatic anomaly detection, and continuous quality and performance monitoring.

The difference from a regular data flow lies in one aspect. Failures do not stop as log entries but instead return data to the party that produced it. Data cleaning work has indeed consumed the largest portion of analysts' time, and automated gates are a way to shift some of that burden onto the system.

Four principles that lock in data layers

Presidential Regulation Number 39 of 2019 concerning One Data Indonesia, effective since June 17, 2019, locks in four requirements that data must meet before it can be shared:

Data Standards, which are the standards underlying specific data, including concepts, definitions, classifications, measurements, and units.
Metadata, which are structured and standardized descriptions that characterize the data and facilitate its search and management.
Data Interoperability, which is the ability of data to be shared between interacting electronic systems. The requirements are consistency in syntax, structure, and semantics, and it must be stored in an open format that can be read by electronic systems.
Reference Codes and Master Data, which are identifiers containing characters as unique data identity references.

These four points constitute a completed data contract. Teams designing DataOps architecture in Indonesia do not need to create them from scratch.

Four stages, and checks that return data

The One Data Indonesia booklet published by the One Data Indonesia Secretariat at the Central Level details the flow into four stages: Data Planning, Data Collection, Data Checking, and Data Dissemination.

The third stage is the least imitated in the private sector. Walidata checks data from Data Producers against the four principles mentioned earlier. If it does not comply, the data is returned for correction, rather than being forwarded with footnotes. Specifically for Priority Data, there is a second check by Data Supervisors according to their fields: statistical data by the Central Statistics Agency, state financial data by the Ministry of Finance, and geospatial data by the Geospatial Information Agency.

In DataOps terms, this is a multi-tier quality gate with clear accountability at each level. The dissemination stage also requires a release schedule and updates to be available for users, making data freshness monitoring not an additional feature but part of the output.

Architecture here means documentation, not a drawing on a whiteboard

The Data List is compiled based on the SPBE Architecture. It is based on Presidential Regulation Number 95 of 2018 concerning the Electronic-Based Government System, effective October 5, 2018, followed by Presidential Regulation Number 132 of 2022 concerning the National SPBE Architecture, effective December 20, 2022. Perpres 132 contains five sections: policy direction and strategy, framework, Architecture References, Architecture Domains, and strategic initiatives. The document is 91 pages long, with the main body only on pages 1 to 5 and the rest as appendices.

Two rarely mentioned figures

Presidential Regulation Number 82 of 2023, effective December 18, 2023, establishes Priority SPBE Applications. One of its categories is applications that are already operational or will be developed with a minimum of 200,000 users or target users. Progress on its implementation is reported to the President every 4 months.

More recently and less discussed, Presidential Regulation Number 83 of 2025, effective July 29, 2025, establishes the Government Digital Transformation Acceleration Committee. This committee is a non-structural body that operates under and is accountable to the President, based on the legal framework of Perpres 95/2018 and Perpres 82/2023.

For data teams, everything boils down to one practical step. Before choosing orchestration tools, first check whether your flow output meets data standards, has standardized metadata, and is stored in an open format. Technical capabilities such as SQL and Python determine how quickly the flow is built, but those three requirements determine whether the output can be used by others. Tools can be replaced, but those requirements cannot.

DataOps Architecture: Four Data Principles That Are Now Mandatory in Indonesia

DataOps places quality gates within the flow, not at the end

Four principles that lock in data layers

Four stages, and checks that return data

Architecture here means documentation, not a drawing on a whiteboard

Two rarely mentioned figures

Sources