Metadata-Version: 2.4
Name: ostia
Version: 0.0.1
Summary: Topology-aware GPU data infrastructure: fabric, columnar exchange, distributed runtime and query engine. Early development.
Keywords: gpu,cuda,rdma,ucx,data-shuffle,distributed-systems,query-engine
Author: Alireza Shateri
Author-email: Alireza Shateri <alirezashateri7@gmail.com>
License-Expression: Apache-2.0
License-File: LICENSE
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: Operating System :: POSIX :: Linux
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: System :: Distributed Computing
Classifier: Topic :: Database :: Database Engines/Servers
Requires-Python: >=3.10
Project-URL: Documentation, https://github.com/OstiaHQ/ostia/tree/main/docs
Project-URL: Homepage, https://github.com/OstiaHQ/ostia
Project-URL: Issues, https://github.com/OstiaHQ/ostia/issues
Project-URL: Source, https://github.com/OstiaHQ/ostia
Description-Content-Type: text/markdown

# Ostia

Topology-aware GPU data infrastructure. Ostia discovers the hardware it runs on (NVLink, NVSwitch, PCIe, NUMA, NICs), measures links, and moves columnar data between GPUs over the best available paths. It is built in layers:

| Package | Layer | Role |
| --- | --- | --- |
| `ostia-fabric` | 1a | Topology discovery, transports, memory registration, byte transfers |
| `ostia-exchange` | 1b | Columnar shuffle, broadcast and gather with flow control and a planner |
| `ostia-runtime` | 2 | Workers, scheduling, fault tolerance, job submission |
| `ostia-query` | 3 | SQL and dataframe queries on GPUs |

**Status: early development.** This release contains no functionality yet; `pip install ostia` will install the full stack once the first components are released.

- Source, design documents and roadmap: https://github.com/OstiaHQ/ostia
- License: Apache-2.0
