使用 Homebrew 安装 vineyard

项目链接

• 主页
• 仓库

命令和别名

• vineyard-graph-loader
• vineyardd

项目历史

The public GitHub repository was created in October 2020, and the project entered the Cloud Native Computing Foundation sandbox on April 28, 2021. CNCF annual reviews describe the original maintainer base as Alibaba-led, with early focus on high-level distributed objects such as tensors, tables, graphs, and streams, plus polyglot SDKs for C++, Python, and Java.

By the 2023 CNCF review, Vineyard had expanded its language and workflow surface to include Python, Go, Rust, Java, Kubernetes integration, workflow-engine integrations, and an academic paper, 'Vineyard: Optimizing Data Sharing in Data-Intensive Analytics,' published at SIGMOD 2023. The project presented itself as a cloud-native object store for complex AI and analytics pipelines where intermediate data exchange is a bottleneck.

采用历史

CNCF review documents report early adoption and evaluation in GraphScope, Mars-related workflows, the ESRF BLISS environment, and PingAn research platforms. The 2023 review says PingAn Tech had moved Vineyard into production for dataset sharing and management, while StartDT/Qidianyun was evaluating it for Python-centric data processing pipelines.

The same CNCF reviews show community growth from the sandbox entry period into 2023, including a reported move from roughly 26 to 40 contributors and from 600-plus to about 750 GitHub stars between the 2022 and 2023 annual reviews. Those figures are self-reported by the project in CNCF governance material rather than independent market-share data.

使用方式

In practice, Vineyard is used as a shared-memory/object-store layer between steps in a data workflow. A producer writes distributed dataframes, tensors, graphs, streams, or other registered object types into Vineyard; consumers in other systems retrieve those objects by id and operate on them without the usual copy-heavy detour through external storage.

Its package-nerd niche is cloud-native data plumbing: the command-line and daemon pieces matter less as user-facing utilities than as the runtime substrate behind Python APIs, Kubernetes operators, scheduler plugins, workflow integrations, and data-system connectors.

时间线

• 2020-10: The v6d-io/v6d repository was created on GitHub.
• 2021-04-28: Vineyard was accepted as a CNCF sandbox project.
• 2022: CNCF annual review described adoption in GraphScope, Mars-related workflows, ESRF BLISS testing, and PingAn testing.
• 2023-06: The Vineyard SIGMOD paper was published with DOI 10.1145/3589780.
• 2023: CNCF annual review described PingAn production use, StartDT evaluation, and expanded workflow/Kubernetes integrations.

Related projects

• Vineyard is closely related to GraphScope, Mars, PyData-style dataframe/tensor workflows, Kubernetes, Kubeflow Pipelines, Airflow, Flyte, Spark, Hive, and the CNCF data/batch ecosystem.

来源

• https://api.github.com/repos/v6d-io/v6d
• https://github.com/cncf/toc/blob/main/reviews/2022-vineyard-annual.md
• https://github.com/cncf/toc/blob/main/reviews/sandbox/2023/2023-vineyard-annual.md
• https://github.com/v6d-io/v6d
• https://github.com/v6d-io/v6d/blob/main/CITATION.cff