/ Gaming Site Can Scale to 250,000 Requests per Second and Improve Player Experience

“Every second counts for a player waiting to place a bet, and by using in-memory OLTP in SQL Server 2014 we provide a faster-loading site and a faster overall experience, so players can place more bets and play games more smoothly.”

Rick Kutschera, Manager of Database Engineering, bwin.party

Online digital entertainment company bwin.party provides online gaming and other offerings to almost 2 million people. Because the organization needed a better ability to support rapid business growth and improve website performance, it deployed an in-memory online transaction processing (OLTP) solution in Microsoft SQL Server 2014. The company can now handle 250,000 requests per second and can offer players a faster, smoother gaming experience. bwin.party also expects to save as much as US$100,000 on hardware costs and is increasing revenue.

This case study is for informational purposes only.
MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS SUMMARY.
Document published March 2014


Business Needs

bwin.party is the world’s largest regulated online gaming company. Hundreds of thousands of people visit the company’s website every day, placing more than 1 million bets on a range of sports, casino games, and poker.

The company was formed from a merger in late 2010 between online gambling giants bwin and Party Gaming. After the merger, bwin.party consolidated its two high-traffic websites, which led to serious scalability issues when players from both sites were directed to a single site that now supported dozens of games from each of the two companies. “We didn’t have the right technology to scale to support all the new users,” says Rick Kutschera, Manager of Database Engineering at bwin.party. “The gaming system only allowed us to handle about 15,000 requests per second. That’s not a lot when you consider our growth rates.” Placing a single bet in a game required approximately 50 batch requests per second. But due to the higher peak loads, the company needed the system’s session state server to support 30,000 batch requests per second.

As a result of the company’s scalability issues, web servers timed out more often, and the site would go down for several minutes at a time. “That lack of availability meant that players couldn’t place bets, and they had a poor experience,” says Kutschera. “Because this business is so competitive, we depend on positive customer experiences. So having any performance issues at all was unacceptable to us, and it also meant that we could lose money because fewer bets were being placed.”

Although the company had recently deployed Microsoft SQL Server 2012 Enterprise software and subsequently installed additional hardware, it still wasn’t enough to address the problem. “We kept trying to optimize the system to make it more scalable, but we really needed a completely new solution to help us do that better.”

Solution

After discussions with Microsoft, bwin.party deployed an in-memory online transaction processing (OLTP) solution in SQL Server 2014. The solution uses main memory optimization and also uses no-locking/no-latching concurrency control mechanisms to remove bottlenecks that are caused by scaling up.

bwin.party decided to test in-memory OLTP shortly after seeing the technology for the first time, becoming one of the earliest Microsoft customers to use the solution in a production environment. “We hadn’t even seen the software bits yet, but we knew we wanted it,” says Kutschera. “That’s how desperate we were to solve our scalability problem.”

The company expected the solution to double the number of requests per second, from 15,000 to 30,000. “We were amazed during the first test run that we actually hit 150,000 requests per second,” Kutschera says.

Soon after, bwin.party integrated the solution with its existing SQL Server 2012 implementation.

Benefits

Using in-memory OLTP in SQL Server 2014, bwin.party can support close to 20 times the number of current site users and can ensure that players have a faster, better experience. The company is also saving money and growing revenue.

Scales to Handle 250,000 Player Requests per Second

With in-memory OLTP, the bwin.party gaming system can scale to 250,000 requests per second, giving the company the ability to easily handle new players on the website. “With SQL Server 2014, we can support close to 20 times the number of players on our site if we need to,” says Kutschera. “If the United States were to legalize online gambling again, we would triple our number of users in one day and we’d have the right technology in place to do that without any performance issues. I even think we’d still be fine if the entire world legalized online gambling.”

Gives Gamers a Faster, Smoother Playing Experience

On the company’s former website, the standard system response time for players was 50 milliseconds. Using in-memory OLTP, that number has improved to 2–3 milliseconds. “The player experience is faster and much better with SQL Server 2014,” says Kutschera.

With much faster performance, players visiting the bwin.party site no longer have delays when loading pages or playing games. “Every second counts for a player waiting to place a bet, and by using in-memory OLTP in SQL Server 2014 we provide faster-loading site and a faster overall experience, so players can place more bets and play games more smoothly. That helps us provide a stronger customer experience overall.”

Reduces Hardware Costs and Boosts Revenue

bwin.party expects to reduce hardware operating costs when in-memory OLTP in SQL Server 2014 is officially launched. “Once this becomes a more widely used system, it will be maintenance-free for us and that will save us a lot of money in the long run,” says Kutschera. “I’ll be able to run in-memory OLTP on less hardware, which could save us as much as $100,000 a year. Previously, we had to throw hardware at scalability and performance problems. But now I can throw software at those problems.”

Additionally, because the solution supports a higher number of players, bwin.party has seen more bets coming into the system. “By using in-memory OLTP we have definitely seen an increase in players, which translates to more revenue for the company,” says Kutschera. “But the biggest thing has been the scalability. Our system has never been as fast or performed as well as it has with SQL Server 2014. This is the solution we were waiting for, and it’s a winning bet for us.”

This case study is for informational purposes only.
MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS SUMMARY.
Document published March 2014