Scaling your Eon Mode database

One of the strengths of an Eon Mode database is its ability to grow or shrink to meet your workload demands. You can add nodes to and remove nodes from your database to meet changing workload demands. For an overview of why you would scale your database and how it affects queries, see Elasticity.

Scaling up your database by starting stopped nodes

The easiest way to scale up your database is to start any stopped nodes:

• To start individual nodes, see Stopping and starting nodes on MC.

• To start an entire subcluster that has been stopped, see Starting and stopping subclusters.

Scaling up your database by adding nodes

If you do not have stopped nodes in your database, or the stopped nodes are not in the subclusters where you want to add new nodes, then you can add new nodes to the database. In supported environments, you can use the MC to provision and add new nodes to your database in a single step. See Viewing and managing your cluster for more information.

You can also manually add new nodes:

• For databases running on AWS, see Add nodes to a running cluster on the cloud.

• For databases running on-premises, or in other environments:

  1. Provision the hardware for your new nodes. Be sure to follow the steps in Before you install Vertica to verify your hardware is configured correctly.

  2. Follow the steps in Adding hosts to a cluster to add the new nodes to the database cluster.

  3. Add the node to your database by following the steps in Adding New Nodes to a Subcluster.

Controlling how Vertica uses your new nodes

New nodes can improve your database's performance in one of two ways:

• Increase the query throughput (the number of queries your database processes at the same time).

• Increase individual query performance (how fast each query runs).

See Elasticity for details on these performance improvements. You control how the new nodes improve your database's performance by choosing what subclusters you add them to. The following topics explain how to use scaling to improve throughput and query performance.