Catalog Details

• CATEGORY

  scaling

• CREATED BY

  Shubham Pandey

• UPDATED AT

  July 08, 2024

• VERSION

  0.0.1

Pattern Snapshot

Related Patterns

scaling

Pod Resource Request

MESHERY4a23

What this pattern does:

This design outlines a Kubernetes architecture tailored for online serving workloads that require GPU acceleration. This design is optimized for Google Kubernetes Engine (GKE), leveraging a single GPU instance to enhance computational performance for machine learning inference, real-time analytics, or other GPU-intensive tasks.

Caveats and Consideration:

Continuous monitoring and optimization of GPU utilization and workload distribution are necessary to maintain optimal performance and avoid resource contention among Pods sharing GPU resources.

Compatibility:

Recent Discussions with "meshery" Tag

• Jun 22 | Newcomers call experience and Star Meshery Sudhanshu Dasgupta
• Jul 05 | So many connected connections Innocentrda
• Jul 03 | Meshery Development Meeting | July 3rd 2024 Vivek Vishal
• Jun 26 | Meshery Development Meeting | June 26th 2024 Yash Sharma
• Jun 27 | Meshery Build and Release Meeting | June 27th 2024 Yash Sharma
• Jun 26 | [Building Meshery with WSL2] Error: invalid memory address or nil pointer dereference (solved) sujal shah
• Jun 22 | Should relationship schemas be the source of truth when defining relationships? Innocentrda
• Jun 18 | Accidently pushed my commit to running PR Aviral Asthana
• Jun 17 | Error while setting up Meshery Dev Environment in Windows Aditya Chari S
• Jun 15 | Where I can find the code of Layer5 Cloud ui? Aviral Asthana