DB Structure for saving form data for dynamically created forms - mysql

Question

I am considering a use case wherein every user of the application can create N number of forms with n different form fields of different data types. Each user can then receive form data in their respective forms. This data can then be updated, edited, searched, sorted etc.
I wonder, how do I design the DB architecture for this in MySQL.
My initial approach was to save form structure in serialized form in one table with form_id as primary key. Another table, data shall hold the records with columns: form_id, record_id, order, value. Here, order will be number denoting the position of the field in the form structure. value column shall have the value for that field. This approach caused me to have 10 rows (for a form with 10 fields) for 1 set of records. Also, I don't think such a query can be written to search records for a particular form.
I did think of using mongodb for this use case, wherein form structure shall be stored in an array and within that array, all records for that particulsr form shall be stored. I have never ever used mongodb, but I guess mongodb has some sort of restriction on size, if I store documents within a document. So, what is the best way to do this using MySQL (or any other DB perfect for such a use case)

Answer 1

Each mongoDB document is restricted to 16mb in size. If you needed each document to be bigger then you would need to adapt your design appropriately. MongoDb does also include GridFS which allows you to break larger files into chunks.
In your example, you could store the form fields as an array in mongoDB but you could also store them as an embedded document. Whichever way you choose to do it, your schema design in mongoDB needs to match your data access patterns.
It's easy to add new key:value pairs to mongoDB documents through your application. Pseudocode would be:
if user adds form field
db.collection.update({'id':1},{$addToSet {'newField':'value'}})
I know your question was about MySQL structure but I thought it worth showing how it could work in mongoDB (obviously would need to be adapted to your chosen language).