Scale-Out MPP Architecture
The speed, scale and rapid growth of today’s information exceed what even the largest of supercomputers could ever handle alone, which is why Massively Parallel Processing (MPP) systems are becoming the norm. Vertica delivers a simple, yet highly robust and scalable MPP solution for the masses with linear scaling and native high availability on industry standard hardware.
No Single Point of Failure
Unlike other MPP Analytic platforms, which are prone to failure and difficult to manage, Vertica has no leader nodes, and therefore no single point of failure. Any node in a Vertica cluster is capable of initiating loads or queries, and will evenly distribute the workload to other nodes when it makes sense to do so. Workload distribution is automatic, no DBA or user intervention required.
Like everything else in the Vertica platform, the MPP layer was an early and deliberate design decision, present in every release of Vertica from inception. This means every other component in Vertica is fully aware of the distributed processing capability, and our platform makes full use of this. Many traditional RDBMS vendors are attempting to bolt on MPP elements to their legacy SMP platforms; however, they are realizing it is much more difficult than just distributing the data.
Vertica’s scale-out MPP architecture leverages off-the-shelf, industry-standard compute and storage devices (servers with disks) to effectively manage scalable deployments at a fraction of the operational and hardware cost of proprietary appliance platforms. Vertica can both scale out and scale up in the event customers wish to deploy Vertica on a shared storage array with a larger host.
Vertica is Fully Elastic
Vertica clusters grow incrementally, by simply adding servers and/or storage automatically and on the fly. Vertica makes it operationally simple to increase storage capacity and increase performance by adding compute and storage resources without taking Vertica offline.