请求-答复数据库模型

Question

I'm wondering about simple database model based on the interaction of requests and replies. The one request can have only one answer. There can be made another request based on the previous answer. It is needed to be able to iterate the all history - from the first request to the last reply. I've done a sample model:

Requests:            Reply:
| int id       |     | int id      |
| int parentID |     | String text |
| String text  |
| int answerID |

How can I improve it and make it more usable?

Answer 1

I'm assuming a request can generate multiple replies.

Request
-------
Request ID
Request timestamp
Request text

Request ID is the primary (clustering) key. It's an auto-incrementing integer or long.

You also define a non-unique index on request timestamp, descending.

Reply
-----
Reply ID
Request ID
Reply timestamp
Reply text

Reply ID is the primary (clustering) key. It's an auto-incrementing integer or long.

You also define a non-unique index on (Request ID, Reply timestamp descending). You may also define a non-unique index on (Reply timestamp descending, Request ID). The first index is for all the replies for a request. The second index is for all the replies in a time period (week , month).