dmlc/dgl

Serving DGL models in production

Open

#4,442 opened on Aug 21, 2022

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Python (12,665 stars) (2,928 forks)batch import
feature requesthelp wanted

Description

Can you share the best practices for serving DGL models in production? (which of the frameworks is preferred/fully supported - torch serve , TensorFlow serving , Kserve or anything kubeflow based , Nvidia Triton .) There are very few resources on the same.

Contributor guide

Tech stack
pythonpytorch
Domain
machine learningbackend
Issue type
research
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Familiarity with DGLKnowledge of model serving
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
10
Research direction
This issue is a general question about best practices for serving DGL models in production. It does not have a specific actionable task. The asker mentions TorchServe, TensorFlow Serving, KServe, and Nvidia Triton. To address this, one could compile a list of supported serving frameworks from DGL's documentation (e.g., https://github.com/dmlc/dgl), compare their performance and community support, and propose a recommended setup. The issue has been idle since 2022 with only one comment, so any response should reference current state.
Serving DGL models in production · dmlc/dgl#4442 | Good First Issue