Clob & JSON parsing vs dedicated table for the nodes in JSON

We are currently evaluating various data layer designs for a project.

We have JSON data (not more than few megs & less than 500 nodes all inclusive) coming in from various sources. We have finalized on 2 approaches. Also we expect not more than 500 concurrent requests.

1. Store the JSON as CLOB in DB, and python json module to parse and extract required fields used by the web application.
2. Design table for the JSON nodes and store it in a dedicated oracle table and retrieve for use with the web application

Need any 3rd degree view/suggestions, before our POC and performance testing.

Thanks for your help. I know the question has broader scope but I think it is specific enough to be considered here.

UPDATE:

1. We use Oracle 11g
2. Requirement is not for reporting but for displaying content on the Web app
3. The web app is expected to be have 10k-30k requests /day, and we would be parsing JSON for every request. So that's 10k-30k times/day.
   4. This is just one part of the bigger web app.

What we are trying to establish is, is it worth having 500 odd columns (many tables) vs JSON parsing which can be handled with Python dicts which are faster anyway.