SQL Query Acceleration for Massively Big Data Sets | SQream

SQL Query Acceleration

In today’s fast-paced environment, organizations need near-real time insights to be able to respond to the changing market. But when running complex queries on massive data sets, query latency with traditional systems can span many minutes to hours. SQream reduces latency to seconds or minutes, resulting in always-updated interactive dashboards.

SQream conforms with the ANSI-92 SQL standard, and adds useful capabilities like window functions, regular expressions and more. By converting SQL queries into clever, highly parallelizable relational algebra operations, SQream can rapidly perform complex operations on the massively parallel GPU cores.

Regular Expressions

SQream supports regular expression pattern matching, letting you effectively filter and sift massive data.

Analytic Functions

With a single query, SQream can rank the top earners or best-selling products across several time-ranges, calculate moving averages, and more.

Join on Any Key

SQream rapidly runs complex joins even when data types don’t match – so you can correlate siloed data without normalizing the data series.

Propelling SQL Query Performance

Use Case: SQream vs.
a Leading Data Warehouse

A leading telecom operator decided to profile SQream in comparison with its existing MPP data warehouse, consisting of 40 compute nodes in 5 full racks. The test data included several months’ worth of CDRs, coming in at 1.6 TB per week. The server used was a repurposed HP DL380g9 with a powerful NVIDIA Tesla card.

Scenario 1: Ad-Hoc Query Performance
In these ad-hoc scenarios, SQream DB outperformed the MPP data warehouse by a factor of 5-13.
Scenario 2: Complex Daily Grinder Report
In this test, a 10-step report was generated. The report queries data from 13 different tables and combines them into a single result-table used to identify top usage location, segmented by customer. The report uses advanced SQL features, including window functions. Using the existing MPP data warehouse, this report would typically take around 2-3 hours. SQream DB outperformed the incumbent by a factor of nearly 18, while providing accurate results from raw data, without pre-aggregations and limiting cubes.

Scale Your Big Data Analytics to Massive Levels

Scale Your Big Data Analytics to Massive Levels

Download this whitepaper to learn how you can dramatically accelerate your existing BI pipeline.