Arize AI Community Icon

Home
Events

Arize AX Releases
Arize AX Support
Arize News
Arize Observe
Discussions
Introductions
Phoenix Support

Powered by Tightknit

Phoenix Support

Deploying Phoenix in Azure Container Apps: Resource Allocation Guide | Arize AI Community

Phoenix Support

Deploying Phoenix in Azure Container Apps: Resource Allocation Guide

·Feb 19, 2025 10:07 AM

Hello everyone, im planning to deploy phoenix as a Container in my Azure Container Apps environment. We are only using the LLM tracing feature and are currently looking at about 3800 LLM / Embedding calls to trace within 24 hours. What resources should i allocate for 1 instance?

4 comments

· Sorted by Oldest

RunLLM
APP
·
For deploying Phoenix in an Azure Container Apps environment, you can start with a configuration similar to what is used internally, which is 2GB RAM and a single CPU. This setup is generally sufficient for handling a moderate workload. However, since you are expecting around 3800 LLM/Embedding calls within 24 hours, you might need to monitor the performance and adjust resources as needed. Phoenix is stateless, so you can scale out by deploying multiple instances backed by a single Postgres database if necessary. This approach allows you to handle increased load by distributing the workload across multiple instances. For more detailed guidance on deployment and scaling, you can refer to the Phoenix Deployment Documentation.Would you like more details on any specific part of this setup?Sources: - https://arize-ai.slack.com/archives/C04R3GXC8HK/p1723828797.781399
💡 Hint: Mention RunLLM in the thread for followups.
Benedikt H.
·
RunLLM thank you very much!
RunLLM
APP
·
I'm here to help with any questions you have about Phoenix, especially regarding its deployment, configuration, and usage for LLM tracing and other features. If you have more questions or need further assistance, feel free to ask!
💡 Hint: Mention RunLLM in the thread for followups.
Xander S.
·
That’s a good starting place