Sharding in layman

A common question I get asked is, "What is Sharding?". Being the complex concept in MongoDB, One of my favorite ways to explain Sharding in plain language is to use an analogy of a wardrobe.

Our clothes are divided and organised in a wardrobe based on how we can access them. A drawer may contain separate compartments for shirts, t-shirts, trousers, and suits, for example. Alternatively, we may keep the matching garments in individual compartments, just like the matching Pants, Shirts, Ties & Belts. It may be preferable for some to keep the same cloth in the same compartment as well.?

The criteria for division are determined by how we gain access to it. In sharding, this is known as a "Shard Key."

If the clothes are arranged by type (shirt, t-shirt, trousers), the person arranging them will know which cloth is in each compartment. In compartment 1, there may be t-shirts, in compartment 2, there may be shirts, in compartment 3, there may be trousers, and so on. These are called "Shards". An individual who knows which type of cloth can be found in a specific compartment is known as a "config server".

If we need a t-shirt, we go straight to the corresponding compartment and get it. That's what we call a "targeted query". Suppose that the distribution is based on the type of cloth, and you would like to search for all red coloured cloth, you would have to search in every compartment; this is an example of a "distributed query".

You can learn more about sharding in M103 course. Feel free to post your questions in our community forum.