Vertica is one of the fastest and efficient analytic databases. It can be installed on cloud or on-premise giving you the ability to be cloud independent. Vertica has a community edition with up to 1 TB for a 3 nodes cluster. Its main features are:
Vertica stores data in a columnar way compared to RDBMS which stores data as rows.
Analytic databases fact tables are large and contain many columns, and most of the time queries use a small number of columns.
This type of storage reduces the IO dramatically as it does not require fetching all rows from the disk, but just the required columns.
By grouping the data in columns, lots of repetitive values are compressed which leads to a smaller amount of data to scan. Vertica can achieve a 90% compression ratio of the data loaded into it.
Vertica Cluster is a shared-nothing architecture (SN), meaning each node stores its data (sharded) locally and in a copy on another node for high availability (in case node fails).
This architecture enhances queries by running them on all nodes in parallel achieving amazing response time.
In case you need to store more data or increase parallelism add more nodes to the cluster.
Vertica does not use Indexes, it uses projection (like materialized views) to store the data in a way business intelligence tools can query it faster. Tables/projections data distribution in the cluster is the key to have the best performance on Vertica.