How to use and secure Azure OpenAi using Private Endpoints | Full Demo
Vložit
- čas přidán 1. 07. 2023
- If you are thinking about using Azure OpenAi and want to know how to use it and secure it, this video will show you how to connect to the public endpoint and then it will show you how to secure the endpoint using Private Endpoint.
full demo that shows you Cross Vnet connections using Vnet Peering and how the DNS works.
I hope you enjoy this video.
all the code used in this video can be found on my github at
github.com/fdubon/youtube/blo...
Thank you - I have been searching forever to understand if Azure OpenAI goes over the interenet or not. Impossible to find such a fundamental question - thank you a million times
I am glad it was helpful
Good video, very detailed, thanks Freddy! We need more demo vids like this!!!
thanks for watching!!
This is the best tutorial on this subject that I've seen. Thanks!
Glad it has helpful!
Amazing stuff Freddy. Really appreciate you creating this content. I will share this with my customers who are struggling with this particular setup
My pleasure!
Excellent video. Thank you very much. Keep up the great content.
Very, Very well explained!!! Kudos my friend!! Excellent video!!
Glad it was helpful!
You are awesome man, Please do more videos like this
Thanks for watching
perfect, that was really helpful.
Glad it helped!
Freddy, I have no words to describe how helpful this was! Thank you so much for taking the time to create this tutorial. I need to do exactly this on my project but using an Azure (premium) function instead of a VM. The only quesiton I have is about your experience with the 'engine' parameter when making a call to the model. I suspect there is some confusion regarding model name vs deployment name. I saw that in your code the deployment name was commented out and model name was used. Did you have any challenges with that? Are there any bugs? Does deployment name needs to be the same as model name? I sometimes can't get the funciton to see the deployment name and get the 'deployment not found' error. Any recommendations would be much appreciated.
Hi @ohaya1, yes this is a bit confusing. I created my script before the deployment was moved to a different area. the model is the one that I call in my script not the deployment name. i have not encountered bugs, the only thing I have encountered is the throttling based on the amount of tokens. The quota takes a while to get it increased in some areas and I am working on using a vector DB to minimize the amount of tokens. I'll keep you uptodate on that. thanks for watching.
I looked into it a bit more and I realized that to avoid any confusion just call your deployment the model name and you will be ok. i hope that helps
@@freddydubon4619 It does, thank you. I was able to solve the issue just recently, it was a bit confusing at first. Thanks so much Freddy - Alex
In my scenario my virtual network is in central US and azure open ai chatgpt-4 is in West US how can connect with this how to create connection private do i need to create a separate virtual network or private endpoints, service endpoints or vnet peering how can you please tell.?
Great video Freddy. Any idea on how to implement this using Terraform?
Thanks.
that is a good question, I do not have a script at the moment that I can share but can work on it
When using private endpoints. Can it connect to different cloud providers i.e AWS via S2S/VPN?
good question!
Well not directly, you can have a VPN or express route from on premises to both providers and then route traffic through that connection. As an alternative, you can setup a VPN connection between the two providers and then you can access the Private endpoint. I hope that makes sense