Data Services: Definition, Value, Lifecycle, Classification and Construction Guidelines

In the past, enterprises relied on traditional integration methods that copied data from one system to another, which became increasingly complex as the number of IT systems grew, causing serious data quality and consistency problems.

Two case studies illustrate the issue: an OLTP contract scenario involving nearly 100 systems and 200 integration relationships, and an OLAP business‑analysis scenario covering 30+ systems across seven domains, both resulting in massive data “movement” and high maintenance costs.

Huawei proposes replacing these fragmented integrations with a data‑service architecture that treats data as a product, delivering it on demand while ensuring reuse, security, and compliance with enterprise standards.

What is a Data Service? According to IEEE‑style definition, a data service is a framework for data distribution and publishing that provides data as a service product to meet real‑time consumer needs, supporting reuse and security.

Data services bring several benefits:

Guarantee a single source of truth, reducing data “movement” and inconsistencies.

Hide technical details from consumers, allowing them to focus on business needs.

Enable rapid, on‑demand data acquisition through subscription.

Support diverse consumption patterns without requiring repeated integration.

Provide fine‑grained access control and usage monitoring for security.

To build data services, Huawei recommends a comprehensive strategy covering the entire service lifecycle: identification & definition, design & implementation, and operation.

1. Service Identification & Definition – Analyze demand, assess reusability, verify entry conditions (owner, metadata, security level, lake‑ingestion), and create an agile delivery plan.

2. Service Design & Implementation – Define service and data contracts, decide granularity based on business, consumption, management and capability characteristics, and establish automated development, testing, and deployment pipelines.

3. Service Change & Retirement – Manage versioning, impact assessment, and de‑registration when usage drops, ensuring consumers are notified and the process is automated.

Data services are classified into two main types:

Data‑Set Services – Provide relatively complete data collections for self‑service analytics; the provider publishes the data set, and consumers retrieve and process it themselves.

Data‑API Services – Respond to event‑driven requests from IT systems, delivering processed results; the provider defines the service contract and handles execution.

Huawei’s “Three‑1s” supply‑chain goal aims to deliver data from request to consumption within 1 day (published services), 1 week (services built on existing lake assets), or 1 month (services built from raw data), supported by clear organizational responsibilities, standardized processes, and dedicated IT platforms.

Overall, the article provides a detailed, practical guide for enterprises to transition from point‑to‑point data integration to a service‑oriented data architecture, improving data quality, agility, and security.