How are left join, inner join, union of mysql can be replicated in redis

Question

I think cross referancing is a good way of learning. I made some research on redis and it seams that as a schameless database redis don't have tables. Instead of tables it has got key-value, key-list, key-set and key-sortedset(zset) pairs.

This kind of key-data pair structure is too unfamiliar for me. I'm more used to mysql, sql styled databases.

I got a sqlFiddle here it demonostrates a simple inner join example.

I offenly take this data from sql and re-model it to a meaningfull JSON object like this:

var posts = {
    users: [{
        userid: 1,
        posts: [{
            header: "header1",
            content: "content of header1"
        },{
            header: "header2",
            content: "content of header2"
        },{
            header: "header3",
            content: "content of header3"
        }]
    },{
        userid: 2,
        posts: [{
            header: "header4",
            content: "content of header4"
        }]
    }]
}

My question is:

How can I mimic this kind of table structure on redis and node.js? Is there a build in redis command for foraign keys? If I have no option but using app level data re-modeling how is the proper way to doing that kind of process.

PS: Cross referancing examples and usefull articals would be apriciated.

EDIT: Is join like operations can be done by SINTER command of Redis?

Answer 1

Redis is a key-value store and not a relational database. As such, there is no native function that would maintain entity relationships. If you wanted it, you'd have to write this into your app.

The closest thing I can think of is what you find in the Mongoose.js module. Mongoose defines a Schema that defines fields used to format and validate documents fetched/saved with MongoDB and also allows for subdocs which could possibly start to mimic relationships between entities.

...but considering that Redis has a record expiration/TTL feature built-in, that records are stored in RAM, that it's a key-value store....it feels a misplaced effort to attempt this. Even Mongoose doesn't provide functionality to enforce referential integrity between collections and limits help to improving record consistency with validation/typing via its Schema.