Vertica Analytic Database performance has been benchmarked in a wide variety of customer settings--different industries, applications, data sets and hardware. The following tables contain benchmark results from some of those tests. Different customers measure different qualities in different ways, thus there is no standard format for the results. If you would like to evaluate Vertica yourself, please request a free evaluation.

Notes: This benchmark was run by a global investment firm. They proved that replacing their existing row-oriented OLTP DBMS with Vertica would improve query performance dramatically. Vertica's compression and built-in active data redundancy help eliminate nearly $1 million dollars annually in storage costs. Furthermore, Vertica's transaction architecture allows concurrent "trickle loading" and querying of data, which enables more real-time views of market data.

Notes: Running on a single server costing less than $10,000 (USD), Vertica out-performed this company's existing CDR database platform by a factor of 40x. As they add servers to the Vertica cluster, performance will scale linearly due to Vertica's shared-nothing column architecture.

Notes: Vertica out-performed the existing CDR data management platform by a large factor and now enables this company to keep months of CDR data on-line for analysis rather than just days. Regarding the bulk load performance, Vertica typically outperforms other databases by a factor of 2x to 3x. In this case, however, the load process in the existing system spent a lot of time aggregating summary data; a process Vertica makes obsolete due to its ability to store and rapidly query data at the lowest level of granularity.

Notes: This benchmark shows the scalability advantages of Vertica's shared-nothing architecture. Increasing the number of CPU cores by a factor of 8 (from 1 server with 4 cores to 4 servers with a combined 32 cores) resulted in just over an 8x increase in Vertica performance yielding an overall 206x improvement over the original row-based DBMS system. This advantage enables people to increase the volume of CDR data available for analysis from weeks to years and also increase the number of people analyzing it without expensive "rip and replace" hardware upgrades.