Understanding the Concept of Mid‑Platform (Zhongtai) and Its Types in Enterprise Architecture

1. What Is Not a Mid‑Platform but Should Be Called a Platform

Development traditionally follows a three‑layer architecture: front‑end presentation, middle logic, and back‑end data. The so‑called “mid‑platform” we discuss does not belong to any of these layers.

Middleware, such as distributed component containers (EJB/COM), distributed transaction coordinators, and message queues, formed a separate market, but the mid‑platform concept is distinct from middleware as well.

In the cloud era, SaaS, PaaS, and IaaS are often mislabeled; many label Spark/Hadoop, TensorFlow, IoT gateways, or media processing services as “mid‑platform.” This over‑use is inaccurate because the true mid‑platform we refer to is not a PaaS offering.

2. Why We Need a Mid‑Platform

Enterprise informationization has entered a new era focused on external, not internal, processes. The first decade (1998‑2008) was the content‑driven PC and mobile internet era; the second (2008‑2018) was the transaction era where enterprises cared about sales rather than clicks.

From 2018 onward, the third era emphasizes the need for rapid upstream production, procurement, and R&D to meet downstream consumer orders. This drives enterprises to move from internal, siloed applications to platforms that connect customers, supply chains, and even societal infrastructure.

Consequently, applications must become micro, scenario‑driven, and fragmentary, especially in the mobile‑first world where traffic is highly fragmented.

3. Business Mid‑Platform

Using new retail as an example, the proliferation of channels, payment methods, and consumer sources creates a demand for unified services such as membership, marketing, order, inventory, payment, and credit. These unified services constitute the business mid‑platform, which can be further specialized into finance, HR, supply‑chain, and new‑manufacturing platforms.

4. Application Mid‑Platform

Beyond business‑specific platforms, there are application mid‑platforms that provide enterprise‑level infrastructure: cloud storage, video conferencing, live streaming, IM, multi‑touch interaction bots, aggregated payment, e‑invoicing, e‑contracts, and bank‑enterprise integration. They are not pure technical middleware but also not tied to a single business scenario, so they are grouped as application components forming the application mid‑platform.

5. Technical Mid‑Platform

Technical mid‑platforms differ from generic IaaS or middleware; they are enterprise‑focused technology platforms. Historically, a technical platform was static—once released, its capabilities remained unchanged for years. With data and AI, platforms now evolve continuously, requiring deployment in public or hybrid clouds to benefit from massive, 360° data training.

Deploying technical platforms in private clouds limits their learning and adaptability. Public or dedicated clouds enable continuous model updates, making the platform smarter and better aligned with societal dynamics.

Integration mid‑platforms are the core of technical platforms, providing capabilities such as integrating internal ERP systems, public‑cloud SaaS, and e‑commerce Open APIs, as well as exposing unified APIs for external ecosystem integration.

6. Data Mid‑Platform

A data mid‑platform carries industry master data, user profiles, business models, and algorithms, whereas a data platform refers to foundational technologies like Hadoop, Spark, Flink, Impala, HBase, Flume, Mahout, and Elasticsearch. Confusing the two is misleading.

Warm Reminder

Please scan the QR code to follow the public account and click the original link for more technical resources and articles.