Columnar database management systems play to their strengths when analysing large volumes of data—such as big data. Since hard drive access constitutes the respective bottleneck when reading any database, and this access is more efficient with a column-oriented DBMS, the columnar variant is really effective here.
For transactional applications, the accesses typically work in a different way. In these cases, new information has to be distributed across the entire database. A columnar database would perform this task more slowly than the conventional model.
Another advantage of columnar databases is the possibility of compressing. Data in one column is always of the same type, for example, a string or an integer. Since all entries of a type are found close together, they can be compressed more efficiently.