Database partitioning - Horizontal vs Vertical - Difference between Normalization and Row Splitting?

Question

I am trying to grasp the different concepts of Database Partitioning and this is what I understood of it:

Horizontal Partitioning/Sharding: Spl

Answer 1

The problems with single database arises when it starts getting huge. So it is required to partition it, to reduce search space, so that it can execute required actions faster.There are various partition strategies available eg: horizontal partitioning, vertical partitioning, hash based partitioning, lookup based partitioning. Horizontal, vertical scaling is different concept compare to these strategies.

1. Horizontal partitioning : It splits given table/collection into multiple tables/collections based on some key information which can help in getting right table as horizontal partitioning will have multiple tables on different nodes/machines. eg: region wise users information.

2. Vertical partitioning : It divide columns into multiple parts as mentioned in one of the above answers eg: columns related to user info, likes, comments, friends etc in social networking application.

3. Hash based partitioning : It uses hash function to decide table/node, and take key elements as input in generating hash. If we change number of tables, it requires re arrangement of data which is costly. So there is a problem when you want to add more table/node.

4. Lookup based partitioning : It uses a lookup table which helps in redirecting to different tables/node base on given input fields. We can easily add new table/node in this approach.

Horizontal scaling vs vertical scaling : When we design any application, we need to think of scaling as well. How are we going to handle huge amount of traffic in future? We need to think in terms of memory consumption, latency, cpu usage, fault tolerance, resiliency. Vertical scaling adds more resources eg: cpu, memory to single machine so that it can handle the in coming traffic. But there are limitation with this approach, you can't add more resource than certain limit. Horizontal scaling allow in coming traffic to distribute across multiple nodes. It need to have load balancer at front which basically handle the traffic, and navigate traffic to any one node. Horizontal scaling allow you to add enough number of servers, but you would also need these many nodes.