Real-Time Loading & Querying
Today, information is generated continuously around the world, and as a result, overnight batch windows of loading data are a luxury of the past. Organizations must be able to load information into their analytics platform nonstop, while simultaneously allowing access to the data for rich analytics.
The time-value of information is critically important, and the sooner users have access to data after it is generated the more valuable it is. For retailers this may mean just-in-time promotions and inventory placement. For financial companies this impacts timely trade decisions. For online gaming companies, this serves up a more personalized and engaging gaming experience. This minimal latency is no easy feat as the volume of information from network feeds, user mouse clicks, financial transactions, sensor networks and a growing number of other sources can be overwhelming.
Hybrid In-Memory/On-Disk Architecture
Unlike traditional RDBMS systems that hold burdensome locks on the entire database while data is being loaded, Vertica was designed with a unique time travel transactional model that ensures extremely high query concurrency, while simultaneously loading new data into the system. Vertica is often able to load data up to 10x faster than traditional row-stores as a result of this design. In addition, Vertica was purpose built with a hybrid in-memory/on-disk architecture to ensure near-real-time availability of information. We automatically query the in-memory and on-disk locations during an analytic operation, and return a single result to the requestor.
Vertica’s FlexStore™ technology allows users to control different storage media tiers for each column of data. This allows the Vertica administrator to easily place the most frequently used columns in a faster storage tier, for instance SSD drives or ioDrives from Fusion-io. The placement of data in the optimal tier and location in the database is automatic from the moment it starts being loaded.