Tencent Announces Open‑Source High‑Performance Graph Computing Framework Plato

Tencent announced on the 14th that it is open‑sourcing its high‑performance graph computing framework Plato, marking the fifth major open‑source project released by Tencent within a week.

Compared with other global graph computing frameworks, Plato can handle graphs with billions of nodes, reducing algorithm execution time from days to minutes. It achieves superior performance while requiring far fewer resources – as few as ten servers can complete computations that previously needed hundreds.

Plato’s team leader, Yu Donghai, stated that Plato already supports many core Tencent services, including WeChat, and provides essential support for Tencent’s massive social network graph data. The framework delivers significant business value and, after open‑sourcing, will continue to drive graph computing research and industry collaboration.

In graph computing, a “graph” refers to an abstract data structure representing relationships between objects, not ordinary images. Graph computing enables the integration of heterogeneous data sources into a single graph for analysis, making it crucial for social networks, recommendation systems, network security, text retrieval, and biomedical research.

Plato was developed by Tencent’s internal TGraph team and is named in honor of the philosopher Plato. Tencent Cloud’s big‑data team is currently packaging Plato for public use.

Performance benchmarks show that Plato outperforms the leading open‑source framework Spark GraphX by 1–2 orders of magnitude, cutting computation time from days to minutes and reducing memory consumption by the same factor. It can complete large‑scale graph tasks on a modest cluster of about ten servers.

Plato offers two core capabilities: offline graph computation and graph representation learning at Tencent‑scale data volumes. It natively integrates with resource schedulers such as Kubernetes and YARN and provides interfaces for various file systems.

The framework’s architecture centers on an adaptive graph computing engine that supports multiple computation modes (adaptive, shared‑memory, pipeline) and extensible communication interfaces.

On top of the engine, Plato provides multi‑layered APIs, a library of graph algorithms, and a “graph toolset” for specific business solutions, enabling integration of offline results with other machine‑learning models.

Several algorithms—graph features, node centrality, connectivity, community detection—are already open‑sourced, with more to follow.

Plato’s high performance, scalability, and plug‑in design make it suitable for social networks, recommendation systems, and biomedical applications such as influence ranking, social graph analysis, and protein interaction studies.

Since the 930 architecture adjustment, open‑source collaboration has become a key strategy for Tencent, leading to the recent open‑source releases of TubeMQ, Tencent Kona JDK, TBase, and TKEStack. Tencent now hosts 89 open‑source projects on GitHub with over 1,000 contributors and more than 260,000 stars, ranking among the top companies globally.

Plato open‑source repository: https://github.com/tencent/plato