Sunday, October 24, 2010

Project Noodle: Initial Database Design

Tip! If you are disoriented, you might want to browse more posts about this project

First off: this will change as I go.

Here’s my first pass at the database tables for Project Noodle:

image

It’s definitely not fully normalized (e.g. Interviews.{CriterionValue1,2,3})—some of that’s intentional. Database design is always a balance.

At this stage, I am really interested in two key things:

  1. Does my schema actually capture all the information I think I need to store?
  2. Are the queries I expect to write easy to write?

And so, being the semi-disciplined person I am (or at least when I’m blogging about a recommendation), I will check these two things.

Does my schema cover everything?

In another window I brought up the mockups. Right off the bat, I’m finding a gap in what I need to capture and what my schema supports:

  • I don’t have anywhere to store why, when, or where a candidate is to be interviewed
  • I don’t have anywhere to store a candidates picture, or resume
  • I don’t have anywhere to store a candidates actual offer or hire status

So let’s fix that together. It looks like I need something akin to an “interview master” table which will store the details of a block of interviews. This can’t really go in the candidates table since a single candidate could come in multiple times.

On the other hand, since HR will likely want to schedule interviews in advance, perhaps I could just create an interview record for each user when the interviews are setup. Then, I can group them by date on the Upcoming Interviews screen.

Having been down this road before, I’m opting for a master table. I’ve found that it helps solve a lot of problems that may not be obvious just yet and it’s probably The Right Thing To Do. Here we go:

image

A more enterprisey solution would probably break out the ratings from the Interviews table, move the photo and resume columns from Candidates into a full blown files table, add a Locations table to hold conference rooms, and a CandidateTypes table for values like “SW Coop”.

I’m opting for the much simpler structure above, deferring that refactoring until a significant problem demands that they be made.

Are my obvious queries easy?

I know I want to display a list of interviews with who/when/why/where information. It looks like I can get all that from the InterviewMasters and Candidates tables. It now occurs to me that I don’t have a way to archive interviews. I’ll ignore that for now and come back to it if it becomes an issue.

I also need to be able to CRU interview details (rankings). This should be very simple with the Interviews table and a couple easy joins.

Finally, I need to be able to aggregate interview feedback from many interviewers into a single view. This also looks very straight forward.

It looks like I haven’t gone overboard with the normalization because I don’t see any red flags so far. I’ve also opted to do no history or versioning, which has helped to avoid a lot of complexity with all of the above queries.

Next

Next up is probably some actual screen development and code.