REVIEWER: Richard Miller, CTO, Next-G Communications, Inc.
BACKGROUND: Next-G Communications is a leading provider of debit-card phone services and developer of telecommunications switching systems.
HARDWARE PLATFORMS: Three Intel servers running Linux.
PROBLEM SOLVED: Our company needs to use the right mix of next-generation networks to maximize quality while reducing delivery costs, improving profitability and delivering new products. Our network managers must make fast, intelligent decisions that balance complex tradeoffs between cost, circuit availability and service quality. While we were accustomed to managing real-time switching and rating systems, we could not provide the required real-time analytic capability using our existing databases. The data we needed was scattered among three database systems: 1) real-time embedded switching, routing and rating data optimized for guaranteed response time; 2) millions of call detail records (CDRs) that document each phone call and enhanced services for transactional services; and 3) back-office systems of distributor, carrier and other information for invoicing and cost-reconciliation purposes. Each system was optimized for the required task; none was optimized for ad hoc querying and what-if analysis. Using the Vertica Database, we built a data mart that streams and integrates data from all three databases. It integrates historical and current CDRs and lets network managers analyze them in near real time to choose the right combination of routes, vendors (TDM and VoIP) and products to improve profitability, quality of service and customer satisfaction. We believe that we are the only company able to manage its routing this way.
PRODUCT FUNCTIONALITY: Vertica is optimized for high performance on read-intensive, query-oriented workloads. Verticas performance comes from: 1) A column-oriented architecture that organizes data on disk as columns of values from the same attribute, as opposed to rows of tabular records. When a query needs to access only a few columns of a particular table (the case for our queries), only those columns need to be read from disk. 2) Aggressive use of compression on disk and on data while its being queried/operated on. 3) Multiple sort orders: because data is compressed, Vertica has sufficient room to store each table on disk in multiple sort orders; tables are decomposed into overlapping groups of columns, called projections, and each projection is sorted on a different attribute (or set of attributes). Each projection is optimized for answering queries with predicates on its sort attributes.
STRENGTHS: Verticas architecture makes it very fast and saves significantly on storage. The database has a shared-nothing, grid-based parallel architecture that runs on clusters of commodity servers, which makes it very economical and scalable. If one server dies, we just hot swap it for another in the field; we are not locked into proprietary hardware at an inflated price. Verticas automated database design capabilities reduce the need for database administrators.
WEAKNESSES: The main weakness of the product is that it is new. We were naturally concerned about quality and stability. However, the products newness has been more than offset by the deep experience of company founder and his team.
SELECTION CRITERIA: Our real-time background taught us that we needed a real-time operating system and supporting language for real-time needs and batch-oriented operating systems and languages for our back-office and batch needs. Similarly, we needed a specialized database for our advanced analytical needs.
DELIVERABLES: The database has a standards-compliant SQL interface, so we were able to drop it right into our reporting environment. Vertica supports Python, which makes interfacing to internal and external Web sites simple and standard. We can now do simultaneous querying of current and historical data. We have been able to automate what-if analyses, data mining and other artificial intelligence-enhanced decision-making processes - things that we could never do before. Instead of setting up queries and building indexes (which could take hours or days), we can just depend on Verticas speed; we are no longer afraid to ask a question for fear well miss a deadline waiting for the answer.
VENDOR SUPPORT: Email support is available quickly. The best part of working with Vertica is having access to the field engineers, who are all seasoned database people.
DOCUMENTATION: The manuals are clear.
The Vertica Database
Vertica Systems, Inc.
Three Dundee Park Drive
Andover, MA 01810-3723
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access