3 Very Different Companies Trust Their Customers’ Experiences to Cloud Infrastructure
Blog: Oracle BPM
As Rohit Sharma led India to a 6-wicket win over South Africa during the Cricket World Cup on June 5, millions of fans hit Reliance Jio’s mobile streaming application to tune in and cheer for Team India.
Bursts of viewers ranging from 100,000 to as many as 5 million users would hit the app at the same time, as India’s manic enthusiasm for the sport propelled a total of 20 million subscribers to stream the live match from their mobile phones via Jio, India’s largest mobile telecom company.
Sudden bandwidth spikes like these can blow through a company’s physical server capacity and cause outages and delays that ruin the experience, and send customers straight to a competitor. Because cloud computing infrastructure is elastic, meaning it can scale up for massive workloads, and back down when demand lets up, upgrading to a cloud infrastructure can help businesses provide reliable service without worrying about their server getting overloaded.
Of course, companies could always add more physical servers, but that takes a lot of time, and it gets expensive, fast, requiring more data center space, staff, and electricity. “We have to handle these kinds of bursts, and give our customers the same great experience, no matter how many CPUs or bandwidth servers we need,” says Gaurav Duggal, Jio’s vice president of technology.
Duggal says Jio ran the live cricket streaming app on bare metal servers on Oracle’s Gen 2 Cloud Infrastructure, and it got faster CPU cycles and better IOPS performance than when it ran similar live sporting events on shared virtual machines from other providers. “Even during those peak load bursts, the number of servers we needed on Oracle was half the number we needed on Amazon,” Duggal says.
Delivering great customer experiences on a massive scale isn’t just a luxury afforded by deep-pocket corporations. Because cloud services are delivered virtually on an “as-needed basis,” companies of all sizes can manage ebbing and flowing workloads, without breaking their backs or their budgets.
Below are examples of two more such companies, which share how they’re providing a better customer experience since upgrading to the cloud.
Car-Sharing Goes Green in Budapest
Car-sharing startup GreenGo offered just 45 electric cars to customers in Budapest when it launched its eco-friendly service in 2016. A year later, the company had expanded its fleet to several hundred Volkswagen e-up! vehicles in the city.
Cofounder and managing director Bálint Michaletzky knew the company needed 300 cars in Budapest to reach its profit targets, but what he didn’t know was how the company was going to operate such a fast-growing mobile business.
“We didn’t have the technology infrastructure to remotely track so many cars, open them, lock them up, send invoices, and receive payments,” he says.
GreenGo started running an instance of Oracle Database on a colocated server, but sharing the server with several other companies led to random CPU allocations that would often leave the startup in the lurch. “We’d have customers hitting our app and requesting cars, but the server simply wouldn’t respond,” especially during peak hours, Michaletzky says.
GreenGo now runs its car-sharing platform on Oracle Cloud Infrastructure, starting with a two-server setup. Initially, it had one running two CPUs, and the other with four CPUs. After load checks, however, the company opted to upgrade both servers to four CPUs. “The entire double setup and switchover took us less than a day,” Michaletzky says. “If we were still running in an on-premises environment, this kind of an upgrade would have taken us four weeks or more to order a new server, set it up, and then start the migration all over again.”
The company is expanding into the Czech Republic, where it plans to offer 200 VW e-ups! in Prague by the end of this year. “We’re always optimizing, and may need to upgrade from four CPUs to eight,” Michaletzky says. “Even so, we expect to complete the entire migration in half an hour.”
With one instance running in Frankfurt, and another in London, Michaletzky attributes GreenGo’s nearly perfect service availability to Oracle’s redundant networks and power supplies. When the company was running on-premises, Michaletzky says it had multiple network and power outages, which could last up to seven hours. “Everything is dependent on the server,” Michaletzky says. “If it’s not available, our customers don’t have a service.”
On Oracle Cloud Infrastructure, “if anything happens in Frankfurt, we still have London. This is exactly why we went with Oracle,” he says.
Latency Sinks Customer Service
Adlib’s business involves quickly turning huge volumes of unstructured paper files, emails, or images into organized and searchable data. It does so using a toolkit of optical character recognition, automated data extraction, artificial intelligence, and big data analytics, for what it calls an “intelligent data conversion platform.”
But doing all that requires enormous compute power and storage capacity—capabilities Adlib’s own data center didn’t have.
“One of our customers had to process 40 million documents within a three-month period. It would have been really difficult for us to deploy all of that with physical servers,” says Mike Grainge, Adlib’s vice president of product engineering.
Adlib helps companies in highly regulated sectors, such as oil and gas, pharmaceutical, insurance, and banking, quickly bring new products to market, while meeting strict compliance requirements and deadlines. With its platform, Adlib identifies important information from unstructured data, and then converts it into a format that makes it easy to analyze, through a process that can include running it against machine-learning algorithms to help classify the data. Once the data is converted into a “high-fidelity” asset, Adlib’s customers can then use the document to understand, for example, which product ideas and strategies it should develop to enter new markets. It’s also valuable to quickly identify information that’s needed for audits or regulatory compliance.
After hitting a wall in its on-premises data center, Adlib initially migrated to Google Cloud Platform. But communication latency delays that could hit 6 milliseconds or more between different data center locations caused problems. While that might not sound like much, “this amount of latency drove up costs and the amount of resources we needed to meet specific time-sensitive processing SLAs for our customers,” Grainge says.
Last year, the company switched to Oracle’s Gen 2 Cloud Infrastructure, and the time has dropped to about 1.5 milliseconds. “When you’re talking about millions of customer documents, that significantly reduces the amount of infrastructure we need to meet our customers’ processing requirements,” Grainge says.
IMAGE: DIBYANGSHU SARKAR/GETTY IMAGES
Leave a Comment
You must be logged in to post a comment.
