Vertica Systems has upgraded its columnar database technology with a focus on performance and analytics.
In Vertica Analytic Database 4.0, the company has added new time series and sessionization to bolster the database's analytic capabilities. The time series features help organizations analyze streams of data that occur at irregular time intervals, such as stock quotes or network monitoring information, by adding the ability to standardize the time intervals and therefore make the analysis much easier.
Sessionization, meanwhile, helps Web properties analyze clickstream data, which can be made difficult because of the challenge of determining what constitutes one session on a Website, explained Dave Menninger, vice president of product management and marketing at Vertica.
"Because the analysis is so difficult, sessionizing the clickstream data is often based on some very simplistic assumptions," he said. "For example, if a click occurs within 30 seconds of the previous click it is part of the same session...[so] not only can the analyses be performed in SQL, but they can also be data driven and specific to each user. Therefore if one user is busy Tweeting, checking Facebook and e-mailing while browsing a Website his or her session characteristics can be significantly different than someone who is doing nothing but browsing the Website."
Around workload management, the company has added the ability for administrators to divide the computing resources into pools and allocate them to specific users and tasks. For example, customer-facing operations can have database resources dedicated and available to ensure that users meet critical requirements while less time-sensitive operations can be given a different set of resources.
Vertica also made changes to its optimizer and execution engine. The parallelization process in 4.0 was dramatically improved, Menninger said, and is now based on the number of cores available in the system.
"In previous versions, parallelization is fundamentally driven by the physical structure of the database," he said. "The second set of enhancements relate to the Vertica's ability to distribute more of the workload equally to the various nodes in the cluster. All MPP databases have a split - some work is done in a distributed fashion and the results of the individual nodes are then compiled or combined in one location within the cluster. ..unlike standard optimizers that determine an optimal single-node plan and then introduce parallelizing operators into it as an afterthought, our patent-pending optimizer algorithms account for data distribution during the join order enumeration phase of the optimizer."
In hopes of expanding its customer base internationally, Vertica 4.0 has been made fully Unicode-capable, meaning it can store and manage data in multiple languages.
"Vertica has dozens of satisfied customers overseas, but Vertica 4.0 will be our flagship for a huge international expansion," said Christopher Lynch, Vertica's CEO, in a statement. "Vertica 4.0 is backed by a rich ecosystem of software, services, server and storage vendors. We intend to fundamentally change the way the world is able to store, access and monetize their data."