The ideal configuration would empower teams of business and data analysts to mine the data residing in existing databases without needing to extract, transform, and load it somewhere else. The team would also be able to make both simple and complex queries, apply different models, run multiple simulations, and more—all without requiring much or any programming. And, because many attempts fail to produce meaningful results, the team would have the ability to tweak existing analyses continuously and run new ones frequently, both of which require getting results quickly.
Now, the advent of GPU-accelerated analytic database solutions supporting user-defined functions makes the massively parallel processing power of GPUs readily accessible to the business analysts who best understand the data and its potential value.
Such user-defined functions leverage application programming interfaces (APIs), “connectors” and other interfaces available in the GPU-powered database. Native API bindings generally support C/C++, Java and Python, and connectors are usually available for the more popular open source and proprietary applications. Of course, these capabilities also give data analysts the ability to create and modify custom algorithms and libraries.
The result is, in effect, the democratization of data science based on the ability to converge AI, machine learning, natural language processing, and BI—all on the same database.
Interactive Location-Based Analytics
Just as most organizations now have a need to process and act on at least some data in real time, most also have a need or desire to integrate location into some data analytics applications. Indeed, in today’s global mobile environment, location data is available from virtually every smartphone and many IoT devices and sensors.
It should come as no surprise that, given the GPU’s roots in graphics processing, an analytics database powered by a GPU is particularly well-equipped to handle the real-time geospatial computations required for interactive location-based analytics. Adding GPU acceleration to a database with a robust, interactive visualization framework makes it possible to ingest, analyze and render billions of geospatial data points in human-interactive time.
The raw geospatial data required for these applications is readily available from a choice of mapping providers, including Google, Bing, Esri, and MapBox. Industry standards make it easy to integrate the data, and with some solutions, users can simply drag and drop analytical applets, data tables and other “widgets” to create and modify customized dashboards.
The benefits of better performance and better price/performance, as well as other advantages of GPU-powered database solutions are now within reach for virtually any organization based on their ability to integrate easily into existing data architectures, and interface with open source, commercial, and/or custom data analytics frameworks.
If your organization is ready to benefit from real-time data analytics, consider implementing a GPU-powered database in a pilot or production application in your own data center. Or, look to the cloud, where GPU instances are now being offered by Amazon, Google, Microsoft, and Nimbix. Either way, the only way to fully appreciate the raw power and real potential of GPU acceleration is to experience it for yourself.