LAS VEGAS: Cloud computing, IBM Watson Analytics and the Internet of Things (IoT) is transforming the way The Weather Company delivers forecasts to 2.3 billion locations.
The firm provides forecasts based on real-time analysis of atmospheric data to locations and organisations across the globe, and has technology demands beyond those of normal enterprises.
The Weather Company started life supplying weather data to broadcasters, but the advent of smartphones and other digital ways to consume forecasts has created a vast demand that the company needs to meet.
David Kenny, chief executive at The Weather Company, told V3 at IBM Insight 2015 in Las Vegas that this has transformed the way the company uses technology.
"The tech journey we're on right now started with the belief that the best way for people to make weather decisions was to deliver [forecasts] through a mobile device, because weather only matters at which location the atmosphere is over," he said.
This means that The Weather Company has to harvest, ingest and analyse vast amounts of data on a daily basis.
"We're constantly ingesting data at all levels of the atmosphere, and then we use algorithms to predict the physics of how that's going to move in order to make that forecast work," said Kenny.
"Then we get the actual data on what happened versus what we predicted would happen, which we can use to go back and constantly learn and improve our algorithms."
The increased demand for forecasts meant that The Weather Company faced the challenge of storing and crunching increasing amounts data in order to rapidly deliver accurate forecasts without encountering crippling infrastructure and compute costs.
Kenny explained that the company needed storage space and access to compute power as the data being collected and used is continuously being ingested and analysed each time it's queried to provide information for a forecast.
"In order to do that cost effectively as we scale up, as we've increased our [data] volumes by about 11-fold to be able to do 26 billion forecasts in a day, we've had to move it all to a cloud-based system," he said.
Enterprises will often opt for a single cloud platform from major providers such as Amazon and Microsoft, but the scale at which The Weather Company operates meant it needed several platforms.
The firm currently uses Amazon Web Services, IBM Cloud and Google's Compute Platform for its cloud-based systems.
The Weather Company created a tool called Grid that enables it to move computational workloads easily between clouds, allowing it to benefit from competition and economies of scale in the cloud market.
Kenny also explained that the company moved from using Hadoop to store data at scale, which used Apache Spark to provide a data query engine, to just using Spark. The reason was to harness the in-memory processing of Spark's framework to enable more efficient data processing and faster forecasting.
"The weather is not like a map. It's constantly moving. So because we've got a very dynamic data set, because we're computing in real time on what's happening today, the speed was most critical to our decision [to use Spark]," said Kenny.
Feeding The Weather Company's data sources is the rise of the IoT and the numerous data collecting devices that create it, which Kenny said gives the company more scope for its forecasting.
Forecast-aiding data can be harvested from connected devices, such as barometric pressure from smartphones and smartwatches, data from connected cars where the activation of windscreen wipers can indicate where rain is falling, and information from the mid-atmosphere gleaned from sensors on the wingtips of airplanes.
"The IoT just gives us so much more observation," said Kenny. "That's just giving us orders of magnitude more data, and we're trying to interpret it."
The final significant part of The Weather Company's technology use is opening up its data sets to third parties to offer forecast data in IBM's analytics portfolio, notably Watson Analytics.
This allows businesses using IBM analytics tools to integrate weather data into the analysis of internal data, thereby gaining an insight into how weather affects their operations and revenue.
The addition of cognitive computing capabilities in Watson Analytics on top of these combined data sets allows patterns to be identified to show how weather conditions affect businesses.
This in turn allows industry-specific decisions to be made, something that normal weather forecasts struggle to facilitate.
"Just being able to learn pattern recognition on unstructured data will help the weather decision be far more clear and dynamic," added Kenny.
Access to The Weather Company's data is also provided through APIs delivered through IBM's Bluemix cloud platform, which allows developers to easily integrate forecast information into apps.
This effectively creates an ecosystem for weather data that applies to specific uses across a wide range of sectors.
Despite its scale, The Weather Company's digital transformation is indicative of a trend that is gaining momentum in the enterprise world, where more businesses want to swap costly infrastructure and static systems for cloud computing and big data analytics.
This can be seen with the likes of General Electric, which is in the middle of an aggressive cloud computing push, and the Williams F1 team, which underwent an IT overhaul to bring in more digital technology.
10nm Cannon Lake Core i3-8121U CPUs make a rare outing with Intel's NUC mini PC
'Notorious' Australian child hacker thought he had executed 'flawless' hack
The former employee says that Tesla fired him for bringing the accusations to management internally
Insecticides based on sulfoxaflor might be as bad for bees as neonicotinoids