Marketing and Advertising Technology
SQream DB uses GPU technology to improve the performance of columnar queries by at least 20x on large data sets.
At the same time reduce the hardware required to perform the query – typically a single 2U server equipped with Nvidia K80 GPU is equivalent to a 42U rack full of servers.
Data science in Ad Tech
Data scientists are true scientists in practice in that they perform experiments to prove which hypothetical models perform the best. A model is iterated several times, on each iteration new parameters are supplied to the model to test accuracy and validity. Each iteration operates on a dataset that is the result set of a query or queries.
A conventional query engine using CPUs alone cannot deliver the result within an acceptable period. The query latency is huge, ranging from many minutes to hours. SQream can execute the same query on the same data set with a latency of seconds to minutes.
Fast discovery of models through reducing query latency of complex queries on large data sets, allows Data Scientists to be more productive and place models into production faster – implicitly reducing costs.
In Ad Tech, when the model is better, the fit is better for the user; the bid price is higher, and a higher price increases the bid/win ratio. The advertiser can spend their budget with a higher likelihood of conversion, and the publisher earns the most money on their inventory.
SQream DB is built from the ground up, to make the best use of available resources, including the revolutionary power of the GPU.
SQream DB combines performance, flexibility (Ad-hoc querying) and ease-of-use, empowering your data science and making discovery insights in your data fast, allowing you to focus on the core of your business, not on the infrastructure.
Standard SQL is used to query the data using standard drivers and tools like JDBC, ODBC, Python, R, Jupyter Notebooks and Spark SQL.
Because SQream uses standard SQL and common language bindings, deep learning technologies that also use GPUs, such as TensorFlow and Theano, work “hand in glove” to reduce the time for modeling and learning experiments.