How to scale like MySpace and Facebook

When a social networking site begins to take off, sometimes its biggest challenge is its own success.

Disaboom Inc. is an online social network aimed at giving individuals with disabilities an outlet to discuss issues ranging from rehabilitation advice to tips on finding jobs. After one full year of operation, the site is currently attracting more than 500,000 unique visitors each month and continuing to grow at a rapid pace.

Harry Brumleve, vice-president of engineering at the Denver, Colo.-based Disaboom, said that the social networking site’s need for scalability and performance in the face of rising membership was a huge concern – especially considering the problems Disaboom faced in the development process.

“My team inherited about 700,000 lines of code when we joined up with Disaboom,” he said. “A consulting company had built the Web site and didn’t meet all the specs needed. They eventually bailed on the contract and hired me to pick up the pieces.”

The problem for Brumleve, who joined Disaboom prior to its launch, was that he needed to finish up the site’s functionality as opposed to making sure the site could scale and perform smoothly. These functionality concerns included making sure the site was compliant with Section 508 of the U.S. Disabilities Act. – enacted to ensure electronic and information technologies are easily accessible to people with disabilities.

“We knew from the get-go when looking at the code we inherited that it wouldn’t scale well,” Brumleve said. “It was architected for about 200 to 300 users. There are 54 million people with disabilities in the U.S. alone and many more who are family members or caregivers. To reach that market, we would hope to have more than 200 to 300 people on our Web site at any given time.”

“You can have the best user experience as far as layout and design,” Brumleve added. “It can be both beautiful and initiative, but if it takes 30 seconds to load, users aren’t going to stay.”

To help increase its Web site performance and reduce the strain on its servers, Disaboom enlisted the help of Vancouver-based Strangeloop Networks Inc. and its AS1000 application scaling tool for Microsoft ASP.Net apps. Installed between web servers and the network, the Strangeloop AS1000 automatically optimizes code, caches Web pages, and reduces payload size to reduce bandwidth needs. Joshua Bixby, president at Strangeloop Networks, said that after implementing AS1000, Disaboom was able to handle big bursts of traffic and dramatically reduce the amount of information it was sending to its data centre.

‘They don’t have the development resources that a MySpace or Facebook, which means they have to be more creative and agile to solve these problems,” he said.

According to Bixby, taking advantage of a non-code-based solution is the best way for any Web-based startup to effectively deal with huge peaks in traffic.

“If you get an article in the New York Times, like Disaboom did, and the traffic goes well beyond your wildest dreams, that isn’t something that you can react to and find ways to optimize the code,” he said. “We’re all about telling our customers to use best practices and write the best code they can. But when you write code, you write it for a specific user group and a specific scale.”

One of the dangers that most startups run into is the impossible task of over engineering their sites to the point that they never see the light of day. Bixby advised Web-based startups to get their code out as quickly as possible, expect that there’s going to be a time when the users start pouring in, and then look at non-code-based solutions to tackle to problem.

Brumleve agreed, saying that Disaboom’s decision to outsource its scalability and performance concerns – which also alleviated the pressure on its development team – has proven to be a beneficial move.

“My guys have a very full plate just handling the business aspects of the blogs and forums we have,” he said referring to his four person development team. “Having them switch gears and find out why the site isn’t running fast would really be a detriment to our business.”

Following the Strangeloop implementation, Brumleve said he’s been extremely pleased with the site’s performance.

“Everything’s really snappy now with our site. The main thing we wanted to solve was payload reduction and we’ve seen that,” he said.

Brumleve added that because a lot of Disaboom’s users spend a great deal of money on medicine and health care, about seven per cent of them are still using dial-up as opposed to cable modem or DSL connections. Using Strangeloop AS1000’s dynamic compression feature, he estimated that the site has reduced its bandwidth needs by about 50 per cent.

Related Download
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center Sponsor: Lenovo
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center
Find out how Hyperconverged systems can help you meet the challenges of the modern IT department. Click here to find out more.
Register Now