[RFC] Support for air gapped mode

[lianhao]

Related to opea-project/GenAIComps#1480, opea-project/GenAIExamples#1405

[lkk12014402]

LGTM. This will benefit network limited users

[yipandeng]

LGTM.

[Copilot]

Pull Request Overview

This PR introduces an RFC to support air-gapped environments for OPEA microservices by outlining the online data types and proposing verification methods.

• Adds a new RFC document outlining air-gap support for microservices.
• Details different online data types and proposed handling methods.

[mkbhanda]

looks good. Awaiting AMD comments.
PS, fixed grammar only because copilot was engaged, wanted to see what it missed!

[lianhao]

Fixed all grammar typo

[mkbhanda]

Have we captured that the tool we build should be executable by the application owner/deployer to ensure OPEA is not violating any re-distribution constraints/license agreements. The tool like our other scripts should take arguments or environment variable, a union of all the environment variables across the application pipeline like HuggingFace token, access token to private container and data repositories etc.

[eero-t]

Have we captured that the tool we build should be executable by the application owner/deployer to ensure OPEA is not violating any re-distribution constraints/license agreements. The tool like our other scripts should take arguments or environment variable, a union of all the environment variables across the application pipeline like HuggingFace token, access token to private container and data repositories etc.

@mkbhanda Models can be downloaded from HF only if owner of given HF token has accepted their licenses.

In general licenses also require that license is included when data is re-distributed, but license info seems to be included only with a subset of models used by OPEA (maybe other models are not downloaded from HF, and/or are public domain?):

$ ls */snapshots/*/LICENSE*
models--Qwen--Qwen2.5-Coder-7B-Instruct/snapshots/0eb6b1ed2d0c4306bc637d09ecef51e59d3dfe05/LICENSE
models--meta-llama--Llama-3.2-3B-Instruct/snapshots/0cb88a4f764b7a12671c53f0838cd831a0843b95/LICENSE.txt
models--meta-llama--Meta-Llama-3-8B-Instruct/snapshots/5f0b02c75b57c5855da9ae460ce51323ea669d8a/LICENSE

[eero-t]

Approved, but consider the remaining 4 suggestions.