#database stuff

okay so i’m taking a server side scripting class, right, and today the teacher pulled up my proposed database structure on the board to talk about how sometimes you’ll make information redundant and “denormalize” a database for the sake of efficiency even though it’s not best practice.

This is because while a client only talks to one server, a server is talking to many different clients and pushing enormous amounts of data to them at all times. This would be fine on its own, but when you bring a database into the picture, you’re constantly calling data from it, and that eats not just bandwidth but also processing power.

So from a strict storage standpoint, you want to have one copy of a piece of information and just call it a whole bunch of times as needed. 

So this was my initial estimate of tumblr, right? its means of storing information is hilariously inefficient. every reblogged post spawns a unique version of the post every time a user reblogs it. why? that’s dumb.

BUT

in order for tumblr to display data on your dashboard, it has to be making frequent calls to its database. so if you’re grabbing a post from a different part of your database, your call has to be more complicated and it starts to bog your server down. Tumblr looks at this big honkin problem and comes up with a solution: fuck normalization, duplicate the post and store it on the same table so we don’t have to make a shit-ton of complicated calls when people start mashing f5.

Tumblr also stands apart from twitter and facebook in its incredible ambitiousness. It isn't just notifying you of posts or displaying lots of small ones, it is a social media site designed to spit out entire blog posts onto your dashboard, lots of them, and with very few limitations. it’s shoving potentially massive amounts of data through its pipes the same way twitter does 280 character tweets with a max of 3 images on them. You can give your hot take instantly, you don’t have to sit down and think out a response. That’s beautiful in a strange way.

it’s also why you can use an extension to edit reblogs. The original post, by the original author isn’t linked or even called in after its placed in your form. you’re just dumping text onto your own blog, so it doesn’t confirm you didn’t change anything because it’d be a waste of time, from an overworked server’s perspective.

In order for posts to reflect the original poster’s edits, tumblr would have to go find the original post and then find and shove in any reblog-added content every time it shows up on someone’s dashboard or blog, so it’s just consolidated into a single static post. You only have to call the database one time instead of an additional time for each comment on the post’s reblog chain. less strain on the server’s aching back.