Difference between Horizontal Scaling and Clustering of Servers

Question

While reading documentation of Cassandra, I came across the term called clustering growth. After reading blogs, I came to know that Clustering is way of grouping of server (Distributed server) via a LAN, to solve the problem, behind it uses the Data Sharding and Partitioning Algorithms. But If we look then in case of Distributed System, where we do horizontal scaling of server. We scale the server horizontally and distribute the load, So we are saying that those server are somehow acheiving the Clustering properties. I basically want to know the difference between Clustering of Server and Replication of Server behind Load Balancer.

I want to know the difference between both of them, Since I knew that clustering is a way for database but I have seen clustered server also. Is Clustering a way of Horizontal scaling or what? Not precisely getting the answer.

Answer 1

In Cassandra we don't tend to scale vertically unless there is a scenario where nodes are under-provisioned. The idea of 'clustering' and 'replication' is built into the very nature of how Cassandra is meant to work.

While you can run Cassandra on a single node, because it is designed as a distributed database, it is most common to have multiple nodes. A group of nodes communicating with each other to make up a distributed database are what we refer to as a cluster. The more nodes you add to a cluster, the more data ownership and workload is spread out, which is where the idea of scaling horizontally comes from.

So, to answer your question, 'clustering' is certainly a way of scaling horizontally when nodes are added to a common cluster to increase throughput. You can also think of a cluster as a logical way to organize data. A Cassandra cluster can have one or more DCs (DataCenters) that are responsible for one or more copies of the data (Replicas) depending on how you define things. I would recommend this quick read for a better understanding: https://cassandra.apache.org/_/cassandra-basics.html