How GPU Databases Are Used to Transform Everything from Smart Cities to the Telco Industry

GPU acceleration is changing the game for fast, location-based analytics. Why? The unique multi-core architecture of GPUs makes it possible to process many computations efficiently and quickly. In fact, the breakout success story this year is around IoT uses cases with GPU databases that “push real-time streaming use cases to the front burner,” according to Ovum’s 2017 Trends to Watch.

GPU databases bring a lot of revolutionary capabilities to IoT data and analytics. First, GPUs take traditional database operations and accelerate them by using thousands of small, efficient cores that are well suited to performing repeated similar instructions in parallel.  Some of the latest GPUs feature over 4,000 cores versus just 16 to 32 cores on a typical CPU-based device. Additionally, GPU cores can crunch data far more efficiently and faster than CPUs, which process data sequentially. These features make GPUs well suited for analyzing massive datasets in real time, particularly for use cases where time and location matter.

In this blog post, we’ll first take a look at a few use cases that incorporate GPU-powered databases for real-time data ingestion. Later in this post, we’ll walk through a demo of how GPU databases are used in the telecommunications industry, and we’ll also take a fun peek into a real-time Twitter feed.

Telco Use Cases

Kinetica, an advanced in-memory analytics database that leverages GPUs for high performance, can be used for a wide variety of telco use cases, such as:

Price Intelligence Monitoring: In the Asia-Pacific region, for example, cell phone providers have two channels: the traditional (reseller) channel, and the modern channel (ATM machine, web, etc.). Typically, a consumer in this region has a prepaid mobile phone that they can recharge – either online or via a retailer. With Kinetica’s geospatial capabilities, telcos can monitor their competitors’ recharge pricing in real time, so they can immediately offer a consumer a specific recharge rate in response to what their competitors in the same area are charging.

Sales/Recharge Channel Profiling & Monitoring: Telecom providers rely on their traditional and modern distribution channels want to be able to understand how effective each channel is. These providers can map their channels to their average revenue per user (ARPU). By using Kinetica to monitor (in real time) where people go to recharge their phones via scratch card or through electronic recharge, they can change their price and tactics. Based on their data, they could decide to invest more in their modern distribution channels versus their traditional resellers.

Network Upgrade: Telcos can also prioritize network upgrades using this streaming data that shows a customer’s type of phone, cell phone usage, and geographic location in real time. By filtering the data, they can easily discover which cell phone tower is used the most, and make upgrade plans accordingly.

Smart Cities/IoT

A smart city is one that uses IoT data to improve aspects of its city operations that are important to its economic vitality, quality of life, safety, and environmental footprint, and to be able to quickly respond to the community’s changing needs.

Trends such as population growth and modernization are putting tremendous strains on the energy generation, transmission, and distribution infrastructure. This makes the ability to discover new value from high volumes of streaming IoT data one of the biggest opportunities in energy today. Energy and public utility companies are utilizing “smart meters” to measure how energy resources such as electricity and natural gas are using in homes and commercial buildings. These metering systems help utilities meet the demand for energy conservation, while also making billing easier for consumers to understand. Homeowners who use smart meters have real-time visibility into their energy consumption and can adjust accordingly, while utilities are better able to meet consumer demand and balance production.

Global trends such as population growth, urbanization, and improved standard of living are forcing city planners and municipalities to deploy smart IoT solutions for traffic and parking management. With a GPU-accelerated database, cities can manage traffic and parking in real time to reduce problems such as congestion and greenhouse emissions.

Let’s take a look at a few demos that show the power of GPUs for high-speed data ingest and real-time data analytics. To view the actual demos, watch the Kinetica/NVIDIA webinar, “GPUs for Accelerating Analytics and Machine Learning.”

Demo: Telco CDR Information

The screen below shows cell phone towers in Indonesia. The actual demo contains about 2,000 records. Each point on the map shows information about a particular tower. The information is then categorized by the number of minutes that a tower has processed, the data volume (in GB) that it has processed, and the total revenue for that tower. You can easily slice and dice the information depending on your needs.

For example, you can zoom in on Jakarta, and look for all of the different calls that originated from Jakarta. You simply click on Jakarta, create a geospatial circle, and within milliseconds, you can view a subset of the data. With this data, you can also perform a timeline-based calculation to view CDR data by day, which is shown in the lower left-hand corner of the screen below.

You can see on the screen below that there is a correlation between the data volumes that a tower has processed and the revenue that a tower generates. For example, in the second bar chart titled Data Volume, you can see that the tower that ends in 743 obviously has a large volume. Similarly, you can see that the revenue is also large. However, if you look deeper, you will see that the fourth tower that ends in 117 does not have a corresponding increase in revenue. This is an anomaly and should be investigated further, as it could be a case of fraud.

Demo: Twitter

The Twitter table below displays 4.126 billion points on a live map where data is streaming in. The actual Kinetica production cluster contains 108 billion points. You can refresh the screen in real time, and the results are displayed in a few milliseconds.

The screen below shows the Asia Pacific region of Tweets. You can easily draw a filter on a geographical region in order to filter the data. In just a few milliseconds, you can drill down from 4.1 billion records to 632 million records.

Kinetica also provides a text search functionality. In this Twitter demo, we can search Tweets for “rain” and “fire” within five words of each other. We were able to drill down from 4.1 billion records to 1 billion records to 9 records in just a few milliseconds.

In this blog post, we hope you learned why GPUs are powering IoT and telco analytics, as well as other use cases that benefit from fast, high-speed data ingest and real-time analytics.

Want to learn more? Watch “GPUs for Accelerating Analytics and Machine Learning,” a webinar in partnership with NVIDIA, featuring speakers Karthik Lalithraj, Principal Solution Architect at Kinetica, and Ettikan Karuppiah from NVIDIA, to learn more about how GPU-powered databases can be used to analyze and visualize data from billions of geospatial objects, on a map, in real time.

Or, find out how Kinetica can work for you. Contact us for a demo today!