<img height="1" width="1" src="https://www.facebook.com/tr?id=1953097804934218&amp;ev=PageView &amp;noscript=1">

Blog

Ars Technica Grows with High Availability Infrastructure

The infrastructure designed, deployed and maintained by ServerCentral has not caused a single minute of downtime for Ars Technica since day one—all the way back in 2008.

The Challenges

Growth
Ars Technica’s popularity continues to grow, so its website must function seamlessly with an extremely large amount of traffic. During highly touted events like Apple Liveblog, arstechnica.com typically grows from 2 million pageviews to 16 million pageviews within two hours. Being able to scale according to irregular fluctuations in demand is critical.

Potential Revenue Loss
Any downtime leads to significant revenue losses for Ars. Advertisers pay per impression, so Ars loses money every second that ads don’t display.

If we go down during something like Apple Liveblog, our users don’t come back. They’d be poached by the competition. Plus, it’s very bad for our reputation.

Lee Aylward, Lead Developer at Ars Technica

The Deployment

In 2012, Ars Technica set out to completely redesign its website, which began with a redesign of its IT infrastructure. Its number one priority was developing a more scalable, robust, and highly available architecture.

We knew we wanted to—needed to—do a high availability infrastructure, but had no idea how to approach it. ServerCentral helped us design the right solution.

Jason Marlin, Technical Director at Ars Technica

The Solution

ServerCentral and Ars determined that a completely managed, high availability architecture was the best approach:

Ars-Ref-Arch 

Instead of pushing us toward the most expensive equipment, ServerCentral’s team was tremendously helpful with arriving at a solution at the cost that we wanted. We hadn’t even considered managed network storage for our VMs until ServerCentral’s engineers suggested it.

Lee Aylward, Lead Developer at Ars Technica

Each component of Ars Technica's fully managed solution is configured for high availability. This ensures that Ars is up 24/7 regardless of equipment failure or maintenance.

Should any active component fail, the system automatically enables running-but-inactive spares and Ars continues to serve their readers and advertisers seamlessly.When a failed component has been replaced or brought online again, ServerCentral reincorporates it into the live environment and reestablishes the automatic failover capability. 

Results

Ars continues to increase visitors and pageviews by more than 20% each year. In addition, the infrastructure designed, deployed, and maintained by ServerCentral has not caused a single minute of downtime for Ars Technica since day one back in 2008.

Cost Reduction

Since having ServerCentral managing our infrastructure, we’ve without a doubt saved time and money. We have cost predictability because they own and manage everything, which our finance team really appreciates.

Jason Marlin, Technical Director at Ars Technica

Support

What I like about ServerCentral’s support is that it has a systematic way of handling things without the cold, robotic approach of your standard ticketing system. I send a request and I’m in their ticketing system, but they add an additional back-and-forth responsiveness anyway. It’s immediate, personal, intelligent, and it’s my favorite part about ServerCentral.

Jason Marlin, Technical Director at Ars Technica

Expertise

If we get a notification from ServerCentral that says ‘we’re looking into a network issue,’ rest assured it’s an actual network engineer working on it who knows a lot more about the problem than we do.

Lee Aylward, Lead Developer at Ars Technica

Convenience

Since we made the move to ServerCentral, we can focus on the programming of our site and user experience. We’re not stressed about infrastructure because it’s running so well.

Lee Aylward, Lead Developer at Ars Technica

Trust

When we get to the point where we have to plan another expansion, we’re doing it with ServerCentral.

Lee Aylward, Lead Developer at Ars Technica

Topics: High Availability